- Email: [email protected]
Riboflavin use as a drug marker in Veterans Administration Cooperative Studies
Riboflavin Use as a Drug Marker in Veterans Administration Cooperative Studies Larry M. Young, Clair M. Haakenson, Kelvin K. Lee, and John P. van Eeckhout Veterans Administration Cooperative Studies Program, Albuquerque, N e w Mexico.
ABSTRACT: Most Veterans Administration (VA) cooperative studies have used only the pill count method to measure and describe patient adherence to a drug regimen. The use of a drug marker is considered when adherence is expected to be a problem in a study, especially if the therapeutic drug or metabolite cannot be measured in the blood or urine of all patients or if reliability of pill counts is open to serious question. In a VA-NHLBI hypertension study using riboflavin as a marker for assessing patient adherence, group data suggest that similar adherence scores can be expected when comparing pill counts and urine tests. However, when an individual patient's adherence was examined by each method at a particular visit, discrepancies were noted. In a VA cooperative study on disulfiram for the treatment of alcoholism, riboflavin used as a marker provided additional information that was needed to assess the adherence of the study population. By employing this second measure of adherence in this study, we were able to obtain at least one measure of adherence at 84% of all clinic visits. If the pill count method had been the sole adherence measure, only 60% of visits would have produced an adherence score. At 65% of clinic visits, the pill count and urine test were in agreement. Results of urine tests taken in the interval between visits were similar to those taken at the clinic visit. Patient cooperation in providing urine specimens or in returning pills to the clinic was slightly associated with positive adherence scores. Although riboflavin has many essential properties of an ideal marker, there are problems associated with its use: (1) patient-to-patient variation in excretion patterns of riboflavin, (2) interference of vitamin preparations and food sources containing riboflavin with normal output concentrations of riboflavin in the urine, and (3) incorporation of riboflavin into a therapeutic drug requiring pharmaceutical company support.
INTRODUCTION In m o s t V A c o o p e r a t i v e studies, w e h a v e u s i n g the pill c o u n t m e t h o d b e e n v e r y successful for m e a s u r i n g a n d d e s c r i b i n g p a t i e n t a d h e r e n c e to the d r u g
Address reprint requests to: Larry M. Young, R. Ph., Clinical Research Pharmacist, Veterans Administration Medical Center, 3350 La Jolla Village Drive, San Diego, Calif. 92161.
Controlled Clinical Trials 5:497-504(1984) © ElsevierSciencePublishing Co., Inc. 1984 52 Vanderbilt Ave., New York, New York 10017
497 0197-2456/84/$03.00
498
L.M. Young et al. regimen. In fact, many studies have higher-than-expected patient adherence scores. Although overestimation can be expected using this method, good adherence in our studies can be attributed to other factors, such as patient selection and counseling [1]. Because of this success with the pill count method, it is the sole method used in many of our studies for measuring adherence. However, the justification for using the pill count method alone is usually strongly supported by our prior experience with the drug and pill-taking habits of the study population. Occasionally in the VA Cooperative Studies Program, a study requires more than one adherence measurement. In these studies, physicians on planning committees advise us of adherence problems, particularly with patients returning unused medication for a pill count. For these studies, a second measurement is selected, usually a test for the drug, or its metabolite, in the blo.od or urine. If this second measurement is complicated by a placebo treatment group, a drug marker would be considered. However, this may not be the only reason or situation in which a drug marker is considered. One should evaluate the disadvantages and advantages of using a drug marker before choosing this adherence measurement, especially when other methods are equally suited for the situation. Since 1964, five VA cooperative studies have used riboflavin as a drug marker for assessing patient adherence to a drug regimen [2-6]. In fact, riboflavin is the only drug marker that has been used in VA cooperative studies. A few cooperative studies have considered using other drug markers as adherence measurements but decided not to prior to finalizing the protocol.
RIBOFLAVIN AS A MARKER
Hobby and Deuschle reported that when riboflavin is incorporated in adequate amounts in tablets or capsules of a therapeutic drug, it may be detected easily in the urine of treated patients for at least 15 to 18 hours after ingestion [7]. From the data available, they concluded that riboflavin concentrations above 2.4 ~g/ml of urine in patients receiving 0.8 mg of riboflavin per kilogram of body weight may be accepted as evidence that the therapeutic drug has been ingested within the previous 18- to 24-hour period, whereas a concentration below 1.5 ~g/ml of urine may be accepted as almost certain evidence that the therapeutic drug has not been ingested. Normal output of riboflavin in the urine, when no vitamin supplements were ingested, rarely exceeded 1.5 ~g/ml. These findings highlight the feasibility of riboflavin as a drug marker in clinical research. In addition, riboflavin has many other properties that make it an ideal drug marker. Table 1 displays the advantages and disadvantages of riboflavin as a drug marker. PAST EXPERIENCE
Riboflavin was first used as a marker more than 15 years ago by the VA Cooperative Study Group on Antihypertensive Agents. In these hypertension studies, 5 mg of riboflavin was incorporated into each study agent. Depending on the dosage regimen, each patient could consume 5-15 mg of riboflavin
Riboflavin Use as a Drug Marker
499
Table 1 Advantages and Disadvantages of Riboflavin as a Drug Marker
Advantages 1. No overt pharmacological effects follow oral administration 2. Readily adsorbed and excreted in the urine within a 24-houre period at recommended doses 3. Nontoxic 4. Nonallergenic 5. Easily detected in urine in low concentrations 6. No discernible coloring of urine 7. No unpleasant odor 8. Palatable 9. Stable under storage conditions employed for therapeutic drugs
Disadvantages 1. Present in many foods and is normally excreted in the urine in low and variable amounts 2. Present in most vitamin preparations 3. Excretion patterns vary from person to person 4. Incorporation of compound into therapeutic drugs requires the support of the pharmaceutical company
daily. At each clinic visit, patient adherence was measured in two ways: by counting returned pills and by inspecting a urine sample for a bright yellow fluorescence under an ultraviolet light. The patient was in pill count adherence if the number of pills was within 80%-110% of the expected from the prescribed regimen and days since a last visit. The patient was in urine adherence if his urine sample fluoresced. No patient was accepted into the randomized phase of these trials unless the urine exhibited fluorescence and the pill counts were within this stipulated range for two successive visits during screening. Some basic problems were encountered using riboflavin in this manner [8]. After three clinical trials, the urine test for adherence was discontinued by this study group. The pill counts alone were thought to be a reliable and valid method of assessing patient adherence. The reasons for dropping the marker as an adherence measure were (1) detecting urine fluorescence under an ultraviolet light was sometimes quite difficult for the nurse clinicians, (2) fluorescence of the urine indicated adherence only the day before a clinic visit, and (3) the pill counts did not always agree with the urine tests for an individual patient. The problem of detecting fluorescence using an ultraviolet light proved to be a difficult task for many nurse clinicians. Many times the urine sample had little fluorescence, and this could only be visualized by comparing a riboflavin-free urine sample to the urine sample being tested. This may have resulted from the low doses of riboflavin being ingested. Some clinic variation in interpretation of results was thought to exist. The second problem was probably .more significant than the first, because many investigators questioned the reliability of a urine test on the day of the clinic visit. Investigators thought patients were more compliant just prior to their clinic visit, and thus a urine test the clinic day would overestimate the true adherence. In our recent study of nonadherence patterns of hypertensive patients, we found no difference in pill count adherence scores the day prior
500
L.M. Young et al. to a clinic visit when compared to overall adherence scores [9]. This study also showed that nonadherence is evenly distributed within a visit period, which suggests that a urine test taken at anytime during a visit would give similar results. The magnitude of the third problem is not well documented in our studies. In fact, the data in Table 2 from the VA-NHLBI Cooperative Study of Mild Hypertension suggest that pill counts and urine fluorescence tests will provide similar results when these two tests are compared; when the criterion for randomization was two successive visits with both pill count adherence and the presence of riboflavin, however, many patients were excluded because these tests did not always agree [5]. Thus, in a few hypertension studies, this adherence criterion for randomization had to be modified during the trial to avoid excluding an excessive number of patients.
PRESENT USE
We are using riboflavin as a marker in a cooperative study of disulfiram for the treatment of alcoholism [6]. In contrast to the hypertension studies, this study f~ses 50 mg of riboflavin (0.8 mg/kg of body weight) in each disulfiram tablet and placebo. The concentration of riboflavin in the urine was determined quantitatively with a photofluorometer at a central location by the method of H o b b y and Deuschle [7]. The patients were scheduled to give a urine specimen weekly for the first 6 months and then every 2 weeks for the next 6 months. Pill counts were performed monthly, and a score greater than 75% was considered positive. A urinary riboflavin concentration above 1.5 ~g/ml was considered a positive indication of adherence. In addition, urine specimens would be analyzed for the presence of diethylamine, a metabolic degradation product of disulfiram, but not all treatment groups could be analyzed for this metabolite. This study is one of the few VA cooperative studies that uses three different measurements of adherence. A drug marker, such as riboflavin, was considered essential since patient adherence with returning unused study medication for pill counts was expected to be a problem and not all treatment groups could be analyzed for a disulfiram metabolite.
ADHERENCE DATA
Of the five VA cooperative studies using riboflavin, the disulfiram study provides the most extensive adherence data to evaluate the usefulness of a Table 2
Comparison of Adherence Estimates From Test of Urine Fluorescence and Pill Counts Within 80%-100% Range a Percent complying by Period Urine fluorescence Pill count Screening 96% 98% First 6 months 88% 89% Whole study 88% 86%
'Adapted from data sourcebook, VA-NHLBICooperativeStudy of Mild Hypertension:Feasibility Trial, Perry, 1975.
501
Riboflavin Use as a Drug Marker
Table 3
C o m p a r i s o n of Clinic Day Riboflavin Urine Test and C o r r e s p o n d i n g Pill C o u n t A d h e r e n c e a
Pill counts adherence levels
Clinic day riboflavin urine test Positive Negative
>75% (positive) 50%-75% (negative) <50% (negative) Total Percent of Visits in Agreement:
333 18 14 365 64.5 percent
164 9 14 187
°Data from VA Cooperative Study on Disulfiram in the Treatment of Alcoholism, K. Lee, Personal Communication, 1982.
riboflavin marker. The pill c o u n t and urine data of this s t u d y were a n a l y z e d to d e t e r m i n e (1) h o w often the pill c o u n t and urine test w e r e in a g r e e m e n t , (2) in w h a t percentage of the visits there was at least one of the two a d h e r e n c e tests p e r f o r m e d , a n d (3) w h e t h e r returning pills and providing urines at clinic visits were associated with higher a d h e r e n c e scores. A major criticism of markers is that they measure a d h e r e n c e in only the 24-hour period immediately preceding the visit, a period in which the patient's a d h e r e n c e m a y be biased b y the u p c o m i n g visit. If a patient r e t u r n e d his bottle for a pill c o u n t at the scheduled clinic visit, we looked at the result of the urine test obtained at the same visit. The data in Table 3 indicate that the pill c o u n t and the urine test were in a g r e e m e n t at 65% of the visits. We a t t e m p t e d to d e t e r m i n e if the urine test results at the clinic d a y visit accurately r e p r e s e n t e d the a d h e r e n c e of the patients b y examining a second urine specimen collected several days prior to the visit. Data in Table 4 indicate that the pill counts were in a g r e e m e n t with the interim urine tests 66% of the time. This suggests that the results of a urine test w o u l d be similar w h e t h e r collected at the clinic visit or at some time b e t w e e n visits. The data in Table 5 s h o w that by e m p l o y i n g two measures of a d h e r e n c e in the disulfiram study, w e obtained at least one m e a s u r e of a d h e r e n c e at 84% of all clinic visits. If the pill c o u n t m e t h o d had b e e n the sole a d h e r e n c e measure, only 60% of visits w o u l d have p r o d u c e d an a d h e r e n c e score. We a t t e m p t e d to d e t e r m i n e if other factors, such as patient cooperation in providing urine specimens or in returning pills to the clinic, were associated
Table 4
C o m p a r i s o n of Interim Riboflavin Urine Test and C o r r e s p o n d i n g Pill C o u n t A d h e r e n c e a
Pill count adherence levels
Interim riboflavin urine test Positive Negative
>75% (positive) 50%-75% (negative) <50% (negative) Total Percent of Visits in Agreement:
495 27 2__22 544 65.6 percent
228 14 2__00 262
~Data from VA Cooperative Study on Disulfiram in the Treatment of Alcoholism, K Lee, Personal Communication, 1982.
502
L.M. Young et al. Table 5
Availability of Urine Specimen and Pill Count Data at 1730 Clinic Day Visitsa Pills returned
Urine specimen obtained Yes
No
Yes 552 (32%) 487 (28%) No 409 (24%) 282 (16%) At least one measure of adherence at 84 percent of visits ~DatafromVA CooperativeStudy on Disulfiramin the Treatmentof Alcoholism,K Lee, Personal Communication, 1982. with positive adherence scores. The data in Table 6 suggest patients that give urine specimens tend to be more pill count-adherent than are patients who do not provide the urines (90% versus 80%). Failure to return pills to the clinic, however, is more suggestive of adherence problems. As suggested by Table 7, about 66% of the patients who returned pills to the clinic had positive urine tests, whereas 49% of the patients who failed to return pills had positive urine tests.
OTHER CONSIDERATIONS Before using riboflavin as a marker, investigators should be aware of problems encountered with its use. . Riboflavin is found in many foods and is normally excreted in the urine in low and variable amounts. Our central adherence laboratory suggested that diets containing varying surpluses of riboflavin will cause the urinary output of riboflavin to vary greatly. A study that looked at the correlation of urinary excretion of riboflavin with dietary intake found that at levels of 1.6-2.55 mg of riboflavin approximately 30% of the dietary riboflavin appeared in the urine [10]. Further increase of the riboflavin level of the diet to 3.55 mg per day raises the urinary output to approximately 50 percent of the amount ingested. Hobby and Deuschle reported that the normal output of riboflavin rarely exceeds 1.5 p,g/ml of urine. However, urinary levels of riboflavin as high as 4 ~g/ml have been reported, and no vitamin supplements were ingested [11]. Table 6
Comparison of Pill Count Adherence and Frequency of Obtaining Urine Specimens at Clinic Visit~ Urine specimen obtained Pill count adherence level Yes No >75% (positive) 50%-75% (negative) K50% (negative) Total
497 (90%) 27 2__88 552 (100%)
391 (80%) 38 5___88 487 (100%)
°DatafromVA CooperativeStudy on Disulfiramin the Treatmentof Alcoholism,K Lee, Personal Communication, 1982.
503
Riboflavin Use as a Drug Marker Table 7
Comparison of Urine Test Results and Frequency of Obtaining Pill Counts at Clinic Visit a Pills returned by patient Urine test Riboflavin Positive Riboflavin Negative Total
Yes
No
365 (66%) 18___7Z 552 (100%)
202 (49%) 20__Z 409 (100%)
°Data fromVA CooperativeStudy on Disulfiramin the Treatmentof Alcoholism,K Lee, Personal Communication, 1982. Additional studies are needed to verify the normal output levels of riboflavin in the urine and the effects of surplus riboflavin diets. 2. Most vitamin preparations contain 2-15 mg of riboflavin per dose. The administration of vitamins containing 15 mg of riboflavin will result in concentrations of riboflavin greater than 1.5 ~g/ml of urine, however, for only 12-18 hours after ingestion of the drug. In a study using a riboflavin marker, a vitamin preparation containing no riboflavin could routinely be prescribed. 3. The pharmaceutical company manufacturing disulfiram experienced tablet discoloration, compressibility, and long disintegration times when modifying the original preparation to incorporate riboflavin. The tablet discoloration was solved by applying an iron oxide-pigmented firm coat, and the tablet compressibility problem was solved by adding a microcrystalized cellulose. One could expect an unwillingness of a pharmaceutical company to get involved with the incorporation of a marker when changes in the basic drug formulations specification result in tablet disintegration problems, and thus bioavailability and bioequivalence changes. No FDA problem should be expected with the use of a riboflavin marker. 4. Excretion patterns after the ingestion of a therapeutic drug containing a riboflavin marker tend to vary from person to person. In the disulfiram study, the average riboflavin concentration was 5-10 p~g/ml. Levels as high as 45 p~g/ml have been recorded. The peak level is usually seen in 8-10 hours. A small-scale pharmacokinetic study of the drug marker before the trial would have been helpful for interpreting these various riboflavin concentrations in the urine.
CONCLUSIONS In many clinical trials, adherence to a drug regimen can be measured successfully with the pill count method alone. However, this method often proves inadequate in clinical trials in which adherence problems are anticipated. Investigators often are unable to obtain a measure of adherence using the pill count method because patients may fail to return pills for counts at clinic visits. Therefore, a second method to assess adherence is recommended. Data from a VA cooperative study suggest that a drug marker such as riboflavin be considered to supplement pill counts. The marker allows adherence to be measured in all treatment groups (active and placebo) and provides adherence data when pill counts are not obtained. In addition, a measure of
504
L.M. Young et al. riboflavin in the urine is a simple, inexpensive, a n d direct m e t h o d of measuring adherence.
REFERENCES 1. Bergman AB, Werner RJ: Failure of children to receive penicillin by mouth. N Engl J Med 268:1334--1338, 1963 2. Veterans Administration Cooperative Study Group on Antihypertensive Agents: Results in patients with diastolic blood pressure averaging 115 through 129 mm Hg. J Am Med Assoc 202:1028-1034, 1967 3. Veterans Administration Cooperative Study Group on Antihypertensive Agents: Multiclinic controlled trial of bethanidine and guanethidine in severe hypertension. Circulation 55(3), 1977 4. Veterans Administration Cooperative Study Group on Antihypertensive Agents: Propranolol in the treatment of essential hypertension. J Am Med Assoc 237:2302-2310, 1977 5. Perry HM: Treatment of mild hypertension--Preliminary results of a two-year feasibility trial. Circ Res 40:1-180, 1977 6. Fuller RK: Disulfiram for the Treatment of Alcoholism. Veterans Administration Cooperative Studies Program Number 107, 1976 7. Hobby GL, Deuschle KW: The use of riboflavin as an indicator of isoniazid ingestion in self-medicated patients. Am Rev Respir Dis 80:415--423, 1959 8. Freis ED: Personal communication. Veterans Administration Medical Center, Washington, DC 9. Tosch TJ, Young LM: A survey of non-adherence patterns in a cooperative study (abstract). Presented at the Third Annual Meeting of the Society for Clinical Trials, Pittsburgh, May 2-5, 1982 10. Hurwitt MK, Harvey CC, Hills OW, Liebert E: Correlation of urinary excretion of riboflavin with dietary intake and symptoms of ariboflavinosis. J Nutr 41:247-264, 1950 11. NeiderhiserD: Personalcommunication. Veterans Administration Medical Center, Cleveland
ABSTRACT: Most Veterans Administration (VA) cooperative studies have used only the pill count method to measure and describe patient adherence to a drug regimen. The use of a drug marker is considered when adherence is expected to be a problem in a study, especially if the therapeutic drug or metabolite cannot be measured in the blood or urine of all patients or if reliability of pill counts is open to serious question. In a VA-NHLBI hypertension study using riboflavin as a marker for assessing patient adherence, group data suggest that similar adherence scores can be expected when comparing pill counts and urine tests. However, when an individual patient's adherence was examined by each method at a particular visit, discrepancies were noted. In a VA cooperative study on disulfiram for the treatment of alcoholism, riboflavin used as a marker provided additional information that was needed to assess the adherence of the study population. By employing this second measure of adherence in this study, we were able to obtain at least one measure of adherence at 84% of all clinic visits. If the pill count method had been the sole adherence measure, only 60% of visits would have produced an adherence score. At 65% of clinic visits, the pill count and urine test were in agreement. Results of urine tests taken in the interval between visits were similar to those taken at the clinic visit. Patient cooperation in providing urine specimens or in returning pills to the clinic was slightly associated with positive adherence scores. Although riboflavin has many essential properties of an ideal marker, there are problems associated with its use: (1) patient-to-patient variation in excretion patterns of riboflavin, (2) interference of vitamin preparations and food sources containing riboflavin with normal output concentrations of riboflavin in the urine, and (3) incorporation of riboflavin into a therapeutic drug requiring pharmaceutical company support.
INTRODUCTION In m o s t V A c o o p e r a t i v e studies, w e h a v e u s i n g the pill c o u n t m e t h o d b e e n v e r y successful for m e a s u r i n g a n d d e s c r i b i n g p a t i e n t a d h e r e n c e to the d r u g
Address reprint requests to: Larry M. Young, R. Ph., Clinical Research Pharmacist, Veterans Administration Medical Center, 3350 La Jolla Village Drive, San Diego, Calif. 92161.
Controlled Clinical Trials 5:497-504(1984) © ElsevierSciencePublishing Co., Inc. 1984 52 Vanderbilt Ave., New York, New York 10017
497 0197-2456/84/$03.00
498
L.M. Young et al. regimen. In fact, many studies have higher-than-expected patient adherence scores. Although overestimation can be expected using this method, good adherence in our studies can be attributed to other factors, such as patient selection and counseling [1]. Because of this success with the pill count method, it is the sole method used in many of our studies for measuring adherence. However, the justification for using the pill count method alone is usually strongly supported by our prior experience with the drug and pill-taking habits of the study population. Occasionally in the VA Cooperative Studies Program, a study requires more than one adherence measurement. In these studies, physicians on planning committees advise us of adherence problems, particularly with patients returning unused medication for a pill count. For these studies, a second measurement is selected, usually a test for the drug, or its metabolite, in the blo.od or urine. If this second measurement is complicated by a placebo treatment group, a drug marker would be considered. However, this may not be the only reason or situation in which a drug marker is considered. One should evaluate the disadvantages and advantages of using a drug marker before choosing this adherence measurement, especially when other methods are equally suited for the situation. Since 1964, five VA cooperative studies have used riboflavin as a drug marker for assessing patient adherence to a drug regimen [2-6]. In fact, riboflavin is the only drug marker that has been used in VA cooperative studies. A few cooperative studies have considered using other drug markers as adherence measurements but decided not to prior to finalizing the protocol.
RIBOFLAVIN AS A MARKER
Hobby and Deuschle reported that when riboflavin is incorporated in adequate amounts in tablets or capsules of a therapeutic drug, it may be detected easily in the urine of treated patients for at least 15 to 18 hours after ingestion [7]. From the data available, they concluded that riboflavin concentrations above 2.4 ~g/ml of urine in patients receiving 0.8 mg of riboflavin per kilogram of body weight may be accepted as evidence that the therapeutic drug has been ingested within the previous 18- to 24-hour period, whereas a concentration below 1.5 ~g/ml of urine may be accepted as almost certain evidence that the therapeutic drug has not been ingested. Normal output of riboflavin in the urine, when no vitamin supplements were ingested, rarely exceeded 1.5 ~g/ml. These findings highlight the feasibility of riboflavin as a drug marker in clinical research. In addition, riboflavin has many other properties that make it an ideal drug marker. Table 1 displays the advantages and disadvantages of riboflavin as a drug marker. PAST EXPERIENCE
Riboflavin was first used as a marker more than 15 years ago by the VA Cooperative Study Group on Antihypertensive Agents. In these hypertension studies, 5 mg of riboflavin was incorporated into each study agent. Depending on the dosage regimen, each patient could consume 5-15 mg of riboflavin
Riboflavin Use as a Drug Marker
499
Table 1 Advantages and Disadvantages of Riboflavin as a Drug Marker
Advantages 1. No overt pharmacological effects follow oral administration 2. Readily adsorbed and excreted in the urine within a 24-houre period at recommended doses 3. Nontoxic 4. Nonallergenic 5. Easily detected in urine in low concentrations 6. No discernible coloring of urine 7. No unpleasant odor 8. Palatable 9. Stable under storage conditions employed for therapeutic drugs
Disadvantages 1. Present in many foods and is normally excreted in the urine in low and variable amounts 2. Present in most vitamin preparations 3. Excretion patterns vary from person to person 4. Incorporation of compound into therapeutic drugs requires the support of the pharmaceutical company
daily. At each clinic visit, patient adherence was measured in two ways: by counting returned pills and by inspecting a urine sample for a bright yellow fluorescence under an ultraviolet light. The patient was in pill count adherence if the number of pills was within 80%-110% of the expected from the prescribed regimen and days since a last visit. The patient was in urine adherence if his urine sample fluoresced. No patient was accepted into the randomized phase of these trials unless the urine exhibited fluorescence and the pill counts were within this stipulated range for two successive visits during screening. Some basic problems were encountered using riboflavin in this manner [8]. After three clinical trials, the urine test for adherence was discontinued by this study group. The pill counts alone were thought to be a reliable and valid method of assessing patient adherence. The reasons for dropping the marker as an adherence measure were (1) detecting urine fluorescence under an ultraviolet light was sometimes quite difficult for the nurse clinicians, (2) fluorescence of the urine indicated adherence only the day before a clinic visit, and (3) the pill counts did not always agree with the urine tests for an individual patient. The problem of detecting fluorescence using an ultraviolet light proved to be a difficult task for many nurse clinicians. Many times the urine sample had little fluorescence, and this could only be visualized by comparing a riboflavin-free urine sample to the urine sample being tested. This may have resulted from the low doses of riboflavin being ingested. Some clinic variation in interpretation of results was thought to exist. The second problem was probably .more significant than the first, because many investigators questioned the reliability of a urine test on the day of the clinic visit. Investigators thought patients were more compliant just prior to their clinic visit, and thus a urine test the clinic day would overestimate the true adherence. In our recent study of nonadherence patterns of hypertensive patients, we found no difference in pill count adherence scores the day prior
500
L.M. Young et al. to a clinic visit when compared to overall adherence scores [9]. This study also showed that nonadherence is evenly distributed within a visit period, which suggests that a urine test taken at anytime during a visit would give similar results. The magnitude of the third problem is not well documented in our studies. In fact, the data in Table 2 from the VA-NHLBI Cooperative Study of Mild Hypertension suggest that pill counts and urine fluorescence tests will provide similar results when these two tests are compared; when the criterion for randomization was two successive visits with both pill count adherence and the presence of riboflavin, however, many patients were excluded because these tests did not always agree [5]. Thus, in a few hypertension studies, this adherence criterion for randomization had to be modified during the trial to avoid excluding an excessive number of patients.
PRESENT USE
We are using riboflavin as a marker in a cooperative study of disulfiram for the treatment of alcoholism [6]. In contrast to the hypertension studies, this study f~ses 50 mg of riboflavin (0.8 mg/kg of body weight) in each disulfiram tablet and placebo. The concentration of riboflavin in the urine was determined quantitatively with a photofluorometer at a central location by the method of H o b b y and Deuschle [7]. The patients were scheduled to give a urine specimen weekly for the first 6 months and then every 2 weeks for the next 6 months. Pill counts were performed monthly, and a score greater than 75% was considered positive. A urinary riboflavin concentration above 1.5 ~g/ml was considered a positive indication of adherence. In addition, urine specimens would be analyzed for the presence of diethylamine, a metabolic degradation product of disulfiram, but not all treatment groups could be analyzed for this metabolite. This study is one of the few VA cooperative studies that uses three different measurements of adherence. A drug marker, such as riboflavin, was considered essential since patient adherence with returning unused study medication for pill counts was expected to be a problem and not all treatment groups could be analyzed for a disulfiram metabolite.
ADHERENCE DATA
Of the five VA cooperative studies using riboflavin, the disulfiram study provides the most extensive adherence data to evaluate the usefulness of a Table 2
Comparison of Adherence Estimates From Test of Urine Fluorescence and Pill Counts Within 80%-100% Range a Percent complying by Period Urine fluorescence Pill count Screening 96% 98% First 6 months 88% 89% Whole study 88% 86%
'Adapted from data sourcebook, VA-NHLBICooperativeStudy of Mild Hypertension:Feasibility Trial, Perry, 1975.
501
Riboflavin Use as a Drug Marker
Table 3
C o m p a r i s o n of Clinic Day Riboflavin Urine Test and C o r r e s p o n d i n g Pill C o u n t A d h e r e n c e a
Pill counts adherence levels
Clinic day riboflavin urine test Positive Negative
>75% (positive) 50%-75% (negative) <50% (negative) Total Percent of Visits in Agreement:
333 18 14 365 64.5 percent
164 9 14 187
°Data from VA Cooperative Study on Disulfiram in the Treatment of Alcoholism, K. Lee, Personal Communication, 1982.
riboflavin marker. The pill c o u n t and urine data of this s t u d y were a n a l y z e d to d e t e r m i n e (1) h o w often the pill c o u n t and urine test w e r e in a g r e e m e n t , (2) in w h a t percentage of the visits there was at least one of the two a d h e r e n c e tests p e r f o r m e d , a n d (3) w h e t h e r returning pills and providing urines at clinic visits were associated with higher a d h e r e n c e scores. A major criticism of markers is that they measure a d h e r e n c e in only the 24-hour period immediately preceding the visit, a period in which the patient's a d h e r e n c e m a y be biased b y the u p c o m i n g visit. If a patient r e t u r n e d his bottle for a pill c o u n t at the scheduled clinic visit, we looked at the result of the urine test obtained at the same visit. The data in Table 3 indicate that the pill c o u n t and the urine test were in a g r e e m e n t at 65% of the visits. We a t t e m p t e d to d e t e r m i n e if the urine test results at the clinic d a y visit accurately r e p r e s e n t e d the a d h e r e n c e of the patients b y examining a second urine specimen collected several days prior to the visit. Data in Table 4 indicate that the pill counts were in a g r e e m e n t with the interim urine tests 66% of the time. This suggests that the results of a urine test w o u l d be similar w h e t h e r collected at the clinic visit or at some time b e t w e e n visits. The data in Table 5 s h o w that by e m p l o y i n g two measures of a d h e r e n c e in the disulfiram study, w e obtained at least one m e a s u r e of a d h e r e n c e at 84% of all clinic visits. If the pill c o u n t m e t h o d had b e e n the sole a d h e r e n c e measure, only 60% of visits w o u l d have p r o d u c e d an a d h e r e n c e score. We a t t e m p t e d to d e t e r m i n e if other factors, such as patient cooperation in providing urine specimens or in returning pills to the clinic, were associated
Table 4
C o m p a r i s o n of Interim Riboflavin Urine Test and C o r r e s p o n d i n g Pill C o u n t A d h e r e n c e a
Pill count adherence levels
Interim riboflavin urine test Positive Negative
>75% (positive) 50%-75% (negative) <50% (negative) Total Percent of Visits in Agreement:
495 27 2__22 544 65.6 percent
228 14 2__00 262
~Data from VA Cooperative Study on Disulfiram in the Treatment of Alcoholism, K Lee, Personal Communication, 1982.
502
L.M. Young et al. Table 5
Availability of Urine Specimen and Pill Count Data at 1730 Clinic Day Visitsa Pills returned
Urine specimen obtained Yes
No
Yes 552 (32%) 487 (28%) No 409 (24%) 282 (16%) At least one measure of adherence at 84 percent of visits ~DatafromVA CooperativeStudy on Disulfiramin the Treatmentof Alcoholism,K Lee, Personal Communication, 1982. with positive adherence scores. The data in Table 6 suggest patients that give urine specimens tend to be more pill count-adherent than are patients who do not provide the urines (90% versus 80%). Failure to return pills to the clinic, however, is more suggestive of adherence problems. As suggested by Table 7, about 66% of the patients who returned pills to the clinic had positive urine tests, whereas 49% of the patients who failed to return pills had positive urine tests.
OTHER CONSIDERATIONS Before using riboflavin as a marker, investigators should be aware of problems encountered with its use. . Riboflavin is found in many foods and is normally excreted in the urine in low and variable amounts. Our central adherence laboratory suggested that diets containing varying surpluses of riboflavin will cause the urinary output of riboflavin to vary greatly. A study that looked at the correlation of urinary excretion of riboflavin with dietary intake found that at levels of 1.6-2.55 mg of riboflavin approximately 30% of the dietary riboflavin appeared in the urine [10]. Further increase of the riboflavin level of the diet to 3.55 mg per day raises the urinary output to approximately 50 percent of the amount ingested. Hobby and Deuschle reported that the normal output of riboflavin rarely exceeds 1.5 p,g/ml of urine. However, urinary levels of riboflavin as high as 4 ~g/ml have been reported, and no vitamin supplements were ingested [11]. Table 6
Comparison of Pill Count Adherence and Frequency of Obtaining Urine Specimens at Clinic Visit~ Urine specimen obtained Pill count adherence level Yes No >75% (positive) 50%-75% (negative) K50% (negative) Total
497 (90%) 27 2__88 552 (100%)
391 (80%) 38 5___88 487 (100%)
°DatafromVA CooperativeStudy on Disulfiramin the Treatmentof Alcoholism,K Lee, Personal Communication, 1982.
503
Riboflavin Use as a Drug Marker Table 7
Comparison of Urine Test Results and Frequency of Obtaining Pill Counts at Clinic Visit a Pills returned by patient Urine test Riboflavin Positive Riboflavin Negative Total
Yes
No
365 (66%) 18___7Z 552 (100%)
202 (49%) 20__Z 409 (100%)
°Data fromVA CooperativeStudy on Disulfiramin the Treatmentof Alcoholism,K Lee, Personal Communication, 1982. Additional studies are needed to verify the normal output levels of riboflavin in the urine and the effects of surplus riboflavin diets. 2. Most vitamin preparations contain 2-15 mg of riboflavin per dose. The administration of vitamins containing 15 mg of riboflavin will result in concentrations of riboflavin greater than 1.5 ~g/ml of urine, however, for only 12-18 hours after ingestion of the drug. In a study using a riboflavin marker, a vitamin preparation containing no riboflavin could routinely be prescribed. 3. The pharmaceutical company manufacturing disulfiram experienced tablet discoloration, compressibility, and long disintegration times when modifying the original preparation to incorporate riboflavin. The tablet discoloration was solved by applying an iron oxide-pigmented firm coat, and the tablet compressibility problem was solved by adding a microcrystalized cellulose. One could expect an unwillingness of a pharmaceutical company to get involved with the incorporation of a marker when changes in the basic drug formulations specification result in tablet disintegration problems, and thus bioavailability and bioequivalence changes. No FDA problem should be expected with the use of a riboflavin marker. 4. Excretion patterns after the ingestion of a therapeutic drug containing a riboflavin marker tend to vary from person to person. In the disulfiram study, the average riboflavin concentration was 5-10 p~g/ml. Levels as high as 45 p~g/ml have been recorded. The peak level is usually seen in 8-10 hours. A small-scale pharmacokinetic study of the drug marker before the trial would have been helpful for interpreting these various riboflavin concentrations in the urine.
CONCLUSIONS In many clinical trials, adherence to a drug regimen can be measured successfully with the pill count method alone. However, this method often proves inadequate in clinical trials in which adherence problems are anticipated. Investigators often are unable to obtain a measure of adherence using the pill count method because patients may fail to return pills for counts at clinic visits. Therefore, a second method to assess adherence is recommended. Data from a VA cooperative study suggest that a drug marker such as riboflavin be considered to supplement pill counts. The marker allows adherence to be measured in all treatment groups (active and placebo) and provides adherence data when pill counts are not obtained. In addition, a measure of
504
L.M. Young et al. riboflavin in the urine is a simple, inexpensive, a n d direct m e t h o d of measuring adherence.
REFERENCES 1. Bergman AB, Werner RJ: Failure of children to receive penicillin by mouth. N Engl J Med 268:1334--1338, 1963 2. Veterans Administration Cooperative Study Group on Antihypertensive Agents: Results in patients with diastolic blood pressure averaging 115 through 129 mm Hg. J Am Med Assoc 202:1028-1034, 1967 3. Veterans Administration Cooperative Study Group on Antihypertensive Agents: Multiclinic controlled trial of bethanidine and guanethidine in severe hypertension. Circulation 55(3), 1977 4. Veterans Administration Cooperative Study Group on Antihypertensive Agents: Propranolol in the treatment of essential hypertension. J Am Med Assoc 237:2302-2310, 1977 5. Perry HM: Treatment of mild hypertension--Preliminary results of a two-year feasibility trial. Circ Res 40:1-180, 1977 6. Fuller RK: Disulfiram for the Treatment of Alcoholism. Veterans Administration Cooperative Studies Program Number 107, 1976 7. Hobby GL, Deuschle KW: The use of riboflavin as an indicator of isoniazid ingestion in self-medicated patients. Am Rev Respir Dis 80:415--423, 1959 8. Freis ED: Personal communication. Veterans Administration Medical Center, Washington, DC 9. Tosch TJ, Young LM: A survey of non-adherence patterns in a cooperative study (abstract). Presented at the Third Annual Meeting of the Society for Clinical Trials, Pittsburgh, May 2-5, 1982 10. Hurwitt MK, Harvey CC, Hills OW, Liebert E: Correlation of urinary excretion of riboflavin with dietary intake and symptoms of ariboflavinosis. J Nutr 41:247-264, 1950 11. NeiderhiserD: Personalcommunication. Veterans Administration Medical Center, Cleveland