Interpretation of Bioavailability Data by Practitioners 6. E. SCHUMACHER. Pharm.D., Ph.D.
ODAY and for the near future, a t least T u n t i l “formularies” are created by state, national, and other health funding agencies and until assurance of bioequivalence is provided by the manufacturers of drug products, practitioners have very little to assist them in their selection of quality dosage forms. To help them in selecting the proper drugs, practitioners make both direct and indirect judgments about the bioavailability of the drug products they prescribe and dispense : ( 1 ) direct judgments based on an evaluation of promotional bioavailability data distributed to them, data which are at present variable in quality and scanty in quantity; and ( 2 ) indirect judgments based on the assessment of reports in the professional literature, the counsel provided by experts, and a perception of the general quality of dosage forms provided by individual drug manufacturers.
Method To gain insight into the ability of practitioners to make direct judgments on bioavailability, we have recently conducted and reported on a study of the bioavailability knowledge of physicians and pharFrom the College of Pharmacy and Allied Health Professions, Wayne State University, Detroit, Mich. 48202. Presented at a Symposium on Bioavailability and Clinical Pharmacokinetics, held at the Fifth Annual Meeting of the American College of Clinical Pharmacology in Philadelphia, Penn., on April 30, 1976. 554
from the SAGE Social Science Collections. All Rights Reserved
Detroit, Mich.
macists.’ A test was developed to assess three aspects of the practitioner’s awarenew of bioavailability: (a) ability to evaluate and make clinical decisions using promotional bioavailability data, (b) ability to demonstrate the basic quantitative and qualitative skills needed to make rational judgments about drug product selection, and (c) ability to discriminate those drugs with actual or potential bioa\-ailability problems. The examination included photocopies of actual and complete promotional bioavailability data brochures presently distributed to practitioners. Thc questions mere based on these data, and the responses of the participants were judged against an answer key developed by using a consensus of responses provided by a panel of 11experts in pharmacokinetics and clinical pharmacology.
Results Examinations were distributed to 254 physicians and pharmacists in the Detroit metropolitan area. Eighteen physicians and 68 pharmacists completed the examination. The mean scores of the participants are shown in Table I. The perforniance of these practitioners is not enconraging and suggests some difficulty in interpreting and evaluating promotional bioavailability data. Pharmacists as a group scored significantly higher than physicians, although there was no significant difference in the performance of community pharmacists as a subgroup and The Journal of Clinical Pharmacologp
INTERPRETATION OF BZOAVAZLABZLZTP DATA
physicians. The difference in scores between clinical pharmacists and the expert panelists was not significant. Clinical pharmacists, however, did score significantly higher than hospital pharmacists rho, in turn, performed better than community pharmacists. Pharmacists who reported that they had participated in various continuing education presentations on biopharmaceutics scored higher than pharmacists who did not participate in continuing education presentations. The relationship was examined between the performance on this test and years since the personnel had received their last academic degree. While no correlation was observed for physicians, there was a correlation noted for pharmacists, and this may reflect the inclusion of courses in biopharmaceutics and pharmacokinetics for pharmacy curricula recently. Limitations in study methodology constrain the interpretation of the results : (a) the sample size was small and limited to one geographic area; (b ) sampling was
nonrandom; (c) while various pharmacy practitioner specialties were represented, the only physicians responding were in academic practice and they represented few specialties; (d) relatively few questions were used to assess certain areas of skill; and (e) the examination as a testing instrument may not adequately reflect the capabilities of practitioners. Nonetheless, these results reinforce my contention that practitioners cannot interpret and evaluate promotional bioavailability data well. Further, they are not capable of discriminating between the usefulness of various types of presentation formats. These observations are critical because the examination questions did not require participants to demonstrate a sophisticated and rigorous background in pharmacokinetics; this background is not a realistic prerequisite for contemporary medical practice. Rather, the test questions required professional judgments about the proper selection of drugs. While a quantitative understanding of
TABLE I
Mean Scores and Ranges on the Bioavailability Examination Sample size
Mean score* (per cent)
Range (per cent)
Panelists
11
68 18
82.7 51.8 35.0
50-100
Pharmacists Physicians
9
74.4
40-100
31
25
56.7 56.4
30-80
27 34
44.6 42.4
10-80 10-80
Groups
10-100
10-60
Subgroups : Clinical pharmacists Pharmacists with continuing education Hospital pharmacists Pharmacists without continuing education Community pharmacists
20-80
Mean score based on the consensus of correct answers as supplied by the panelists. Score for individual panelists was the standard (82.7 per cent) obtained by judging their responses against an answer key.
October, 1976
555
SCEUMACHER
pharmacokinetics is valuable for practitioners, it is not required for the judicious interpretation of bioavailability data today. Practitioners are not called upon to make mathematical calculations of product bioavailability. After all, manufacturers are not going to submit promotional or experimental data to practitioners which do not support the bioequivalence of their products to accepted “standards.” The real problem for practitioners today-and the examination results provide some evidence for this concern-is not so much in evaluating data that fail to support bioequivalence but in recognizing errors or omissions in drug study design, assessment, or presentation that are a part of the promotional or clinical data given to practitioners. I n discussing these problems with practitioners, I have divided the potential problems in study design, evaluation, and presentation into three categories2: 1. Pitfalls in interpreting the significance of therapeutic data: (a) bioavailability data do not reflect real range of therapeutic plasma concentrations, urine concentrations, o r body stores; (b) assessments using single dose peak plasma concentration data; (c) single dose data given for drugs used on multiple dosage regimens; (d) in vitro data that imply in vivo performance. 2. Pitfalls of methodology: (a) differing patient populations, assay techniques, and study protocols; (b) incomplete blood sampling interval. 3. Pitfalls in presentation of the data: (a) standards not properly defined; (b) no references to lot numbers of products; (c) methodology citations incomplete.
Pitfalls in Interpreting Significance of Data Since it is impractical to expect that all drug manufacturers are capable of producing bioavailability data for all of the multisource drug products in the next few 556
years, it is necewary to set priorities for acquiring the data. From a practical Viewpoint, varying bioavailability for a multisource dosage form of the same drug entity is much more critical for some categories of drugs than for others. Drugs with narrow therapeutic indices (e.g., the cardiac and aminoglycosides, antiarrhythmics, anticonvulsants, bronchodilators, lithium) are of great concern because there is a small range of acceptable blood and tissue concentrations. On the other hand, drugs with broad therapeutic indices (like many of the antimicrobials) o r drugs whose relationship between body stores and therapeutic response is poorly characterized (like some of the psychotherapeutics) are of lesser concern. From these observations, it is apparent that the clinical significance of potential variations in product bioavailability is a practical consideration today because of the limitations in acquiring and interpreting data. To reinforce this point, it is instructive to compare and speculate on the consequences of the variations in bioavailability among capsules of ampicillin and tablets of digoxin. There is no evidence published t o sup port the contention that a 30 per cent variation in the bioavailability of various brands of ampicillin capsules leads to therapeutic inequivalence. Thus, 1000 mg versus 700 mg daily or 1000 mg versus 1400 mg daily of the same brand of ampicillin could be similar in efficacy. Furthermore, the prescribing information accompanying ampicillin capsules recommends a dose of 1000 mg daily for patients of 20 kg o r more who have infections of the respiratory tract or soft tissue^.^ This same dosage is used for patients of varying weights, which suggests either a broad therapeutic index for the drug, a poor understanding of the relationship between therapeutic response and dose, or both. This is not to suggest that bioequivalence should not be assured for The Journal of Clinical Pharmacology
INTERPRETATION 03’BIOAVAILABILITP DATA
all brands of ampicillin, for surely it should be; but it does suggest that a 30 per cent variation in bioavailability may not be of real consequence considering our current knowledge, contemporary drug usage, and the relatively nonspecific dosage recommendations for ampicillin. On the other hand, by using some of the dose-response data from the literature on digoxin,’.6 it is possible to predict the therapeutic consequence of a 30 per cent change in bioavailability for patients stabilized on digoxin who normally achieve daily peak body stores of approximately 10 micrograms/kg of lean body weight. Changing brands for these patients leads to a 30 per cent increase in absorption, which is an unwitting bonus unknown t o the physician, pharmacist, and patient. Unfortunately, this change can be estimated to increase the patients’ risk of a glycoside-induced arrhythmia by 100 per cent and the general risk of an adverse drug reaction by 150 per cent.5 On the other hand, a 30 per cent decrease in absorption leads to an estimated 50 per cent or greater chance of not receiving therapeutic benefit from the drug. This hypothetical example serves as a prototype for other drugs with a narrow therapeutic index. These practical considerations of therapeutic performance must be exercised Khen interpreting (a) bioavailability data obtained by comparing single-dose peak blood level data, (b) single-dose data for drugs commonly used on multiple-dose regimens, and (c) in vitro data that imply in vivo performance. The assessment of bioavailability data which compare products solely on the basis of peak blood level data derived from a single dose may be misleading unless the data are known to be characteristic of the therapeutic performance; variations in absorption rate (often reflected in differences in peak blood level) among competing products of the same drug are usually not of clinical October, 1976
consequence. The use of single-dose bioavailability data is appropriate, and certainly realistic in view of the high costs of conducting such studies, as long as the studies have been performed using proper methods, as discussed below. Nonetheless, there is no question that steady-state data, achieved only upon multiple-dose administration, are more informative and indicative of the actual clinical performance of drugs. Practitioners should be skeptical, however, of promotional bioavailability data that use in vitro data alone to imply in vivo performance. Using in vitro data as the sole test of eficacy is only acceptable when the in vitro test (usually for dissolution) is known to reasonably predict steady-state in vivo results. This has been shown, for example, for digoxin by some investigators.6
Pitfalls in Methodology It is important for practitioners to verify that the same study methods were used in determining the bioavailability of both the test and reference products described in the promotional data. Dittert and DiSantoe and Chun and SeitzlO have described the pitfalls involved in interpreting results from inadequately conducted studies. They have also provided insight into the potential for inaccurately interpreting data when identical study conditions are not maintained for both the test and reference products. Also, practitioners should assure that a crossover technique has been used in conducting bioavailability studies. Even properly designed studies performed by different manufacturers must be evaluated with caution. For example, the promotional bioavailability data supplied for erythromycin tablets by Abbott and Upjohn appear to represent appropriately designed and conducted studies, both comparing the other’s product, yet arriving at different results because the study protocols were not the same.2*11J2 557
SCHUMACHER
Another frequently encountered pitfall in methodology is the presentation of bioavailability data based on an incomplete blood sampling interval. Evaluations made on the basis of incomplete sampling may bias results in favor of the most rapidly absorbed dosage form, but it may not be that the differences in the extent of absorption o r that the differences in absorption rate have clinical significance. Under certain conditions, it is possible to conduct abbreviated bioavailability tests, but this procedure is only appropriate when it has been demonstrated that the shorter sampling interval correlates with the results obtained from a complete sampling interval. Yet, the practitioner is generally unaware of whether the abbreviated sampling that is used in some promotional bioavailability presentations is predictive of more extensive sampling. Some rules of thumb are helpful to determine this. The comparison of the test and reference products is probably valid if ( a ) a crossover technique is employed, (b) other aspects of methodology are appropriate, (c) the blood level curves for the two products are superimposable or, if not, the curves for the two products are declining in parallel, and (d) the concentration of the last sample is one sixth or less of the peak concentration reported. The comparison of the products may not be valid, however, even if a crossover technique or other appropriate methodology is employed, if either the blood level curves for the two products are not declining in parallel or the last concentration sampled is greater than one fourth of the peak concentration. Pitfalls in Presentation of Data
If practitioners are to correctly interpret data that imply drug product equivalence or inequivalence, then they must be supplied not only with supportive statements of methodology, as I previously suggested, but also with the identity of the 558
reference product used as the basis for comparison. It is not possible to reach a meaningful interpretation of promotional bioavailability data when the test product is compared to some undisclosed reference standard. It is also useful to request lot numbers of the test and reference products used in performing the bioavailability study. Using this information, it is possible to request verification, upon subaquent product purchases, that no formulation changes have occurred that may alter the bioavailability of the product from the lot. tested. Information to be Requested of Manufacturers I n conclusion, I advise practitioners to solicit information from manufacturers in the seven critical categories listed below to assist them in making the most meaningful interpretation of the promotional bioavailability data they receive.2 1. Examine the study protocol. Request a complete disclosure of study group characteristics, methods of conducting the study, and analytical procedures used to evaluate blood or urine concentrations of drug. 2. Identify the reference product. 3. Identify the test and reference product lot numbers. 4. Identify the source of supply. Determine who the manufacturer of dosage form is, and check the possibility of multiple producers supplying the same manufacturer or distributor and that the bioavailability data supplied actually represent the product being purchased. 5. Scrutinize in vitro data. Assess quality by requesting of the manufacturer standard quality control data such as content uniformity, hardness, disintegration and dissolution tests, etc., in addition to a visual evaluation of quality. 6 . Examine therapeutic blood level data. Request information relating blood levels to therapeutic response. This information The Journal of Clinical Pharmacology
INTERPRETATION OF BIOAVAILABILITP DATA
may assist you in the practical interpretation of the blood level bioavailability data. 7. Interpret statistical analysis. Request a statistical interpretation of the significance of the results.
Beferences 1. Tinson, B., and Schumacher, Q.: A study of the bioavailability knowledge of physicians and pharmacists, presented at the Midyear Clinical Meeting of the American Society of Hospital Pharmacists, Washington, D.C., December 10, 1975. 2. Schumacher, G.: Use of bioavailability data by practitioners I: pitfalls in interpreting the data. Amer. J. Hosp. Pharm. 32:839
(1975). 3. Baker, C.: Physicians’ Desk Reference, 30th ed. Medical Economics Co., Oradell, N.J., 197ti. 4. Jelliffe, R., Buell, J., and Kalaba, R.: Reduction of digitalis toxicity by computerassisted glycoside dosage regimens. Ann. Int. Med. 77:891 (1972). 5. .Jelliffe, R., and Brooker, G.: A nomogram for digoxin therapy. Amer. J. Y e d . 57: 63 (1974).
October, 1976
6. Lindenbaum, J., Butler, V., Murphy, J., and
7.
8.
9. 10.
11.
12.
Cresswell, R. : Correlation of digoxin-tablet dissolution-rate with biological availability. Lancet 1 :1215 (1973). Johnson, B., Greer, H., McCrerie, J., Bye, C., and Fowle, A.: Rate of dissolution of digoxin tablets as a predictor of absorption. Lancet 1 :1473 ( 1973). Sham, T., Raymond, K., and Howard, M.: Therapeutic nonequivalence of digoxin tablets in the United Kingdom: correlation with tablet dissolution rate. Brit. Ned. J. 4:763 (1973). Dittert, L., and DiSanto, A.: The bioavailability of drug products. J. Amer. Pharm. Assoc. A-S 13:421 (1973). Chun, A., and Seitz, J.: Drug bioavailability and its utility. J. Amer. Pharm. Assoc. N S 14:407 (1974). The Upjohn Co.: E-mycin Protocol CS #77, promotional bioavailability data distributed upon request by The Upjohn Co., Kalaniazoo, hlich., 1973. Abbott Laboratories. : Protocol for Study Nos. 72-9 and 72-82, promotional bioavailability data distributed upon request by Abbott Laboratories, North Chicago, 111.
559
ODAY and for the near future, a t least T u n t i l “formularies” are created by state, national, and other health funding agencies and until assurance of bioequivalence is provided by the manufacturers of drug products, practitioners have very little to assist them in their selection of quality dosage forms. To help them in selecting the proper drugs, practitioners make both direct and indirect judgments about the bioavailability of the drug products they prescribe and dispense : ( 1 ) direct judgments based on an evaluation of promotional bioavailability data distributed to them, data which are at present variable in quality and scanty in quantity; and ( 2 ) indirect judgments based on the assessment of reports in the professional literature, the counsel provided by experts, and a perception of the general quality of dosage forms provided by individual drug manufacturers.
Method To gain insight into the ability of practitioners to make direct judgments on bioavailability, we have recently conducted and reported on a study of the bioavailability knowledge of physicians and pharFrom the College of Pharmacy and Allied Health Professions, Wayne State University, Detroit, Mich. 48202. Presented at a Symposium on Bioavailability and Clinical Pharmacokinetics, held at the Fifth Annual Meeting of the American College of Clinical Pharmacology in Philadelphia, Penn., on April 30, 1976. 554
from the SAGE Social Science Collections. All Rights Reserved
Detroit, Mich.
macists.’ A test was developed to assess three aspects of the practitioner’s awarenew of bioavailability: (a) ability to evaluate and make clinical decisions using promotional bioavailability data, (b) ability to demonstrate the basic quantitative and qualitative skills needed to make rational judgments about drug product selection, and (c) ability to discriminate those drugs with actual or potential bioa\-ailability problems. The examination included photocopies of actual and complete promotional bioavailability data brochures presently distributed to practitioners. Thc questions mere based on these data, and the responses of the participants were judged against an answer key developed by using a consensus of responses provided by a panel of 11experts in pharmacokinetics and clinical pharmacology.
Results Examinations were distributed to 254 physicians and pharmacists in the Detroit metropolitan area. Eighteen physicians and 68 pharmacists completed the examination. The mean scores of the participants are shown in Table I. The perforniance of these practitioners is not enconraging and suggests some difficulty in interpreting and evaluating promotional bioavailability data. Pharmacists as a group scored significantly higher than physicians, although there was no significant difference in the performance of community pharmacists as a subgroup and The Journal of Clinical Pharmacologp
INTERPRETATION OF BZOAVAZLABZLZTP DATA
physicians. The difference in scores between clinical pharmacists and the expert panelists was not significant. Clinical pharmacists, however, did score significantly higher than hospital pharmacists rho, in turn, performed better than community pharmacists. Pharmacists who reported that they had participated in various continuing education presentations on biopharmaceutics scored higher than pharmacists who did not participate in continuing education presentations. The relationship was examined between the performance on this test and years since the personnel had received their last academic degree. While no correlation was observed for physicians, there was a correlation noted for pharmacists, and this may reflect the inclusion of courses in biopharmaceutics and pharmacokinetics for pharmacy curricula recently. Limitations in study methodology constrain the interpretation of the results : (a) the sample size was small and limited to one geographic area; (b ) sampling was
nonrandom; (c) while various pharmacy practitioner specialties were represented, the only physicians responding were in academic practice and they represented few specialties; (d) relatively few questions were used to assess certain areas of skill; and (e) the examination as a testing instrument may not adequately reflect the capabilities of practitioners. Nonetheless, these results reinforce my contention that practitioners cannot interpret and evaluate promotional bioavailability data well. Further, they are not capable of discriminating between the usefulness of various types of presentation formats. These observations are critical because the examination questions did not require participants to demonstrate a sophisticated and rigorous background in pharmacokinetics; this background is not a realistic prerequisite for contemporary medical practice. Rather, the test questions required professional judgments about the proper selection of drugs. While a quantitative understanding of
TABLE I
Mean Scores and Ranges on the Bioavailability Examination Sample size
Mean score* (per cent)
Range (per cent)
Panelists
11
68 18
82.7 51.8 35.0
50-100
Pharmacists Physicians
9
74.4
40-100
31
25
56.7 56.4
30-80
27 34
44.6 42.4
10-80 10-80
Groups
10-100
10-60
Subgroups : Clinical pharmacists Pharmacists with continuing education Hospital pharmacists Pharmacists without continuing education Community pharmacists
20-80
Mean score based on the consensus of correct answers as supplied by the panelists. Score for individual panelists was the standard (82.7 per cent) obtained by judging their responses against an answer key.
October, 1976
555
SCEUMACHER
pharmacokinetics is valuable for practitioners, it is not required for the judicious interpretation of bioavailability data today. Practitioners are not called upon to make mathematical calculations of product bioavailability. After all, manufacturers are not going to submit promotional or experimental data to practitioners which do not support the bioequivalence of their products to accepted “standards.” The real problem for practitioners today-and the examination results provide some evidence for this concern-is not so much in evaluating data that fail to support bioequivalence but in recognizing errors or omissions in drug study design, assessment, or presentation that are a part of the promotional or clinical data given to practitioners. I n discussing these problems with practitioners, I have divided the potential problems in study design, evaluation, and presentation into three categories2: 1. Pitfalls in interpreting the significance of therapeutic data: (a) bioavailability data do not reflect real range of therapeutic plasma concentrations, urine concentrations, o r body stores; (b) assessments using single dose peak plasma concentration data; (c) single dose data given for drugs used on multiple dosage regimens; (d) in vitro data that imply in vivo performance. 2. Pitfalls of methodology: (a) differing patient populations, assay techniques, and study protocols; (b) incomplete blood sampling interval. 3. Pitfalls in presentation of the data: (a) standards not properly defined; (b) no references to lot numbers of products; (c) methodology citations incomplete.
Pitfalls in Interpreting Significance of Data Since it is impractical to expect that all drug manufacturers are capable of producing bioavailability data for all of the multisource drug products in the next few 556
years, it is necewary to set priorities for acquiring the data. From a practical Viewpoint, varying bioavailability for a multisource dosage form of the same drug entity is much more critical for some categories of drugs than for others. Drugs with narrow therapeutic indices (e.g., the cardiac and aminoglycosides, antiarrhythmics, anticonvulsants, bronchodilators, lithium) are of great concern because there is a small range of acceptable blood and tissue concentrations. On the other hand, drugs with broad therapeutic indices (like many of the antimicrobials) o r drugs whose relationship between body stores and therapeutic response is poorly characterized (like some of the psychotherapeutics) are of lesser concern. From these observations, it is apparent that the clinical significance of potential variations in product bioavailability is a practical consideration today because of the limitations in acquiring and interpreting data. To reinforce this point, it is instructive to compare and speculate on the consequences of the variations in bioavailability among capsules of ampicillin and tablets of digoxin. There is no evidence published t o sup port the contention that a 30 per cent variation in the bioavailability of various brands of ampicillin capsules leads to therapeutic inequivalence. Thus, 1000 mg versus 700 mg daily or 1000 mg versus 1400 mg daily of the same brand of ampicillin could be similar in efficacy. Furthermore, the prescribing information accompanying ampicillin capsules recommends a dose of 1000 mg daily for patients of 20 kg o r more who have infections of the respiratory tract or soft tissue^.^ This same dosage is used for patients of varying weights, which suggests either a broad therapeutic index for the drug, a poor understanding of the relationship between therapeutic response and dose, or both. This is not to suggest that bioequivalence should not be assured for The Journal of Clinical Pharmacology
INTERPRETATION 03’BIOAVAILABILITP DATA
all brands of ampicillin, for surely it should be; but it does suggest that a 30 per cent variation in bioavailability may not be of real consequence considering our current knowledge, contemporary drug usage, and the relatively nonspecific dosage recommendations for ampicillin. On the other hand, by using some of the dose-response data from the literature on digoxin,’.6 it is possible to predict the therapeutic consequence of a 30 per cent change in bioavailability for patients stabilized on digoxin who normally achieve daily peak body stores of approximately 10 micrograms/kg of lean body weight. Changing brands for these patients leads to a 30 per cent increase in absorption, which is an unwitting bonus unknown t o the physician, pharmacist, and patient. Unfortunately, this change can be estimated to increase the patients’ risk of a glycoside-induced arrhythmia by 100 per cent and the general risk of an adverse drug reaction by 150 per cent.5 On the other hand, a 30 per cent decrease in absorption leads to an estimated 50 per cent or greater chance of not receiving therapeutic benefit from the drug. This hypothetical example serves as a prototype for other drugs with a narrow therapeutic index. These practical considerations of therapeutic performance must be exercised Khen interpreting (a) bioavailability data obtained by comparing single-dose peak blood level data, (b) single-dose data for drugs commonly used on multiple-dose regimens, and (c) in vitro data that imply in vivo performance. The assessment of bioavailability data which compare products solely on the basis of peak blood level data derived from a single dose may be misleading unless the data are known to be characteristic of the therapeutic performance; variations in absorption rate (often reflected in differences in peak blood level) among competing products of the same drug are usually not of clinical October, 1976
consequence. The use of single-dose bioavailability data is appropriate, and certainly realistic in view of the high costs of conducting such studies, as long as the studies have been performed using proper methods, as discussed below. Nonetheless, there is no question that steady-state data, achieved only upon multiple-dose administration, are more informative and indicative of the actual clinical performance of drugs. Practitioners should be skeptical, however, of promotional bioavailability data that use in vitro data alone to imply in vivo performance. Using in vitro data as the sole test of eficacy is only acceptable when the in vitro test (usually for dissolution) is known to reasonably predict steady-state in vivo results. This has been shown, for example, for digoxin by some investigators.6
Pitfalls in Methodology It is important for practitioners to verify that the same study methods were used in determining the bioavailability of both the test and reference products described in the promotional data. Dittert and DiSantoe and Chun and SeitzlO have described the pitfalls involved in interpreting results from inadequately conducted studies. They have also provided insight into the potential for inaccurately interpreting data when identical study conditions are not maintained for both the test and reference products. Also, practitioners should assure that a crossover technique has been used in conducting bioavailability studies. Even properly designed studies performed by different manufacturers must be evaluated with caution. For example, the promotional bioavailability data supplied for erythromycin tablets by Abbott and Upjohn appear to represent appropriately designed and conducted studies, both comparing the other’s product, yet arriving at different results because the study protocols were not the same.2*11J2 557
SCHUMACHER
Another frequently encountered pitfall in methodology is the presentation of bioavailability data based on an incomplete blood sampling interval. Evaluations made on the basis of incomplete sampling may bias results in favor of the most rapidly absorbed dosage form, but it may not be that the differences in the extent of absorption o r that the differences in absorption rate have clinical significance. Under certain conditions, it is possible to conduct abbreviated bioavailability tests, but this procedure is only appropriate when it has been demonstrated that the shorter sampling interval correlates with the results obtained from a complete sampling interval. Yet, the practitioner is generally unaware of whether the abbreviated sampling that is used in some promotional bioavailability presentations is predictive of more extensive sampling. Some rules of thumb are helpful to determine this. The comparison of the test and reference products is probably valid if ( a ) a crossover technique is employed, (b) other aspects of methodology are appropriate, (c) the blood level curves for the two products are superimposable or, if not, the curves for the two products are declining in parallel, and (d) the concentration of the last sample is one sixth or less of the peak concentration reported. The comparison of the products may not be valid, however, even if a crossover technique or other appropriate methodology is employed, if either the blood level curves for the two products are not declining in parallel or the last concentration sampled is greater than one fourth of the peak concentration. Pitfalls in Presentation of Data
If practitioners are to correctly interpret data that imply drug product equivalence or inequivalence, then they must be supplied not only with supportive statements of methodology, as I previously suggested, but also with the identity of the 558
reference product used as the basis for comparison. It is not possible to reach a meaningful interpretation of promotional bioavailability data when the test product is compared to some undisclosed reference standard. It is also useful to request lot numbers of the test and reference products used in performing the bioavailability study. Using this information, it is possible to request verification, upon subaquent product purchases, that no formulation changes have occurred that may alter the bioavailability of the product from the lot. tested. Information to be Requested of Manufacturers I n conclusion, I advise practitioners to solicit information from manufacturers in the seven critical categories listed below to assist them in making the most meaningful interpretation of the promotional bioavailability data they receive.2 1. Examine the study protocol. Request a complete disclosure of study group characteristics, methods of conducting the study, and analytical procedures used to evaluate blood or urine concentrations of drug. 2. Identify the reference product. 3. Identify the test and reference product lot numbers. 4. Identify the source of supply. Determine who the manufacturer of dosage form is, and check the possibility of multiple producers supplying the same manufacturer or distributor and that the bioavailability data supplied actually represent the product being purchased. 5. Scrutinize in vitro data. Assess quality by requesting of the manufacturer standard quality control data such as content uniformity, hardness, disintegration and dissolution tests, etc., in addition to a visual evaluation of quality. 6 . Examine therapeutic blood level data. Request information relating blood levels to therapeutic response. This information The Journal of Clinical Pharmacology
INTERPRETATION OF BIOAVAILABILITP DATA
may assist you in the practical interpretation of the blood level bioavailability data. 7. Interpret statistical analysis. Request a statistical interpretation of the significance of the results.
Beferences 1. Tinson, B., and Schumacher, Q.: A study of the bioavailability knowledge of physicians and pharmacists, presented at the Midyear Clinical Meeting of the American Society of Hospital Pharmacists, Washington, D.C., December 10, 1975. 2. Schumacher, G.: Use of bioavailability data by practitioners I: pitfalls in interpreting the data. Amer. J. Hosp. Pharm. 32:839
(1975). 3. Baker, C.: Physicians’ Desk Reference, 30th ed. Medical Economics Co., Oradell, N.J., 197ti. 4. Jelliffe, R., Buell, J., and Kalaba, R.: Reduction of digitalis toxicity by computerassisted glycoside dosage regimens. Ann. Int. Med. 77:891 (1972). 5. .Jelliffe, R., and Brooker, G.: A nomogram for digoxin therapy. Amer. J. Y e d . 57: 63 (1974).
October, 1976
6. Lindenbaum, J., Butler, V., Murphy, J., and
7.
8.
9. 10.
11.
12.
Cresswell, R. : Correlation of digoxin-tablet dissolution-rate with biological availability. Lancet 1 :1215 (1973). Johnson, B., Greer, H., McCrerie, J., Bye, C., and Fowle, A.: Rate of dissolution of digoxin tablets as a predictor of absorption. Lancet 1 :1473 ( 1973). Sham, T., Raymond, K., and Howard, M.: Therapeutic nonequivalence of digoxin tablets in the United Kingdom: correlation with tablet dissolution rate. Brit. Ned. J. 4:763 (1973). Dittert, L., and DiSanto, A.: The bioavailability of drug products. J. Amer. Pharm. Assoc. A-S 13:421 (1973). Chun, A., and Seitz, J.: Drug bioavailability and its utility. J. Amer. Pharm. Assoc. N S 14:407 (1974). The Upjohn Co.: E-mycin Protocol CS #77, promotional bioavailability data distributed upon request by The Upjohn Co., Kalaniazoo, hlich., 1973. Abbott Laboratories. : Protocol for Study Nos. 72-9 and 72-82, promotional bioavailability data distributed upon request by Abbott Laboratories, North Chicago, 111.
559
