CLINICAL PHARMACOLOGY an d
THERAPEUTICS volume 18 number 2
August, 1975
Commentary Major drug-prescribing patterns in general hospitals Prescriptions of major drug classes were surveyed for one year in five teaching and five nonteaching general hospitals . Their overall rates varied over a two- to three-fold range in medicine, surgery, and pediatrics, but were uniform throughout the hospitals. Parenteral fiuids and antibiotics represented over one half of the pharmacotherapy used in every specialty except medicine, where tranquilizers were the mostfrequently prescribed drugs. Prescribing patterns lacked consistency in obstetrics and newborn services, and generally with steroids, vasopressors, and anticoagulants. Significant variations also involved the use offluids in pediatrics, diuretics in surgery, and vasodilators in medicine, These variations were too unpredictable and large to be attributed to specific differences among patient populations. Only two discrepancies were noted between teaching and nonteaching hospitals: tranquilizers were prescribed more often in the former, and vasopressors in the latter.
Jean Sice Cambridge, England Special Commission on Internal Pollution
Drug use in hospitals," clinics," and offices'? has received a great deal of attention during the last decade. These studies were conducted for a variety of reasons, the most urgent of which remain the dangers of medication abuse. Thus, the frequency of mishaps that complicate or prolong an illness has been shown to depend on the number of drugs taken." Complex regimens produce drug interactions," increase interferences with diagnostic tests," and multiply errors of dispensing!! and compliance. 7 Reprint requests 10: Dr. Jean Sice , 5139 Fifth Ave .. Pittsburgh, Pa. 15232.
The severity of adverse reactions, on the other hand, largely depends on the nature of the treatments: parenteral water and electrolytes, antiarrhythmics, tranquilizers, and anticoagulants account for the majority of lifethreatening accidents. 4 Antibiotics, also, can alter the characteristics of infectious diseases. 9 Other considerations involve the time spent in processing drug orders, starting and adjusting intravenous infusions, and the cost of pharmaceuticals." Knowledge of patterns of drug use, moreover, might help phannacology instructors to balance the importance given to the 133
134
Sice
Table I. Mean prescribing rates services 0/ 10 general hospitals
Clinical Pharmacolog y and The rape utics
0/ major drug
classes surveyed in medical and surgical
Patient exposure in per cent 0/ all (N) patients Drug classes
Thyroid and antithyroid agents Oral antidiabetics Vasopressors Miscellaneous anti-infectives Sulfonamides Insulin Antihypertensivcs Vasodilators Steroids Anticoagulants Diuretics Antiarrhythmies Antibiotics Tranquilizers Parenteral water and electrolytes
Medicine = 55,702)
(N
2.08 3.74 1.95 2.19 1.95 5.75 6.70 7.35 7.83 11.3 20.4 23.0 25.5 44.1 31.4
I
(N
Surgery = 87,782)
0.82 1.57 2.02 2.12 3.70 2.00 1.71 1.55 4.26 3.28 6.43 5.26 33.6 27.8
44.4
Drug utilization in per cent 0/ all (N) treatments Medicine (N = 108,713)
1.07* 1.92 1.00* 1.12* 1.00* 2.94 3.43 3.76 4.01 5.77 10.4 11.8 13.0 22.6 16.1
I
Surgery (N = 123.336)
0.59* 1.12* 1.43 1.51 2.63 1.42 1.21* 1.10* 3.03 2.34 4.57 3.74 23.9 19.8 31.6
*This class of drugs was one uf all those that consritutcd 4'K of the agents prescribed in this specialty.
various classes of drugs in educational programs." Most of the current information on prescribing, however, is derived from studies of general medical services in university hospitals. The pharmacotherapy practiced in the other specialties and in nonteaching institutions still is conjectural. The present survey, therefore , was undertaken in order to obtain this information. Its first objective was to determine prescribing rates of major drug classes in medicine, surgery, pediatrics , obstetries , and newborn in general hospitals. The second was to see if the use of these drugs conformed to some pattern, and to find if there might be consistent differences between teaching and nonteaching institutions. Methods
The data of this survey covered the entire calendar year 1972. They were obtained from the Commission on Professional and Hospital Activities, Ann Arbor, Michigan. But aIl analyses, interpretations, and conclusions based on these data are solely those of the author; the Commission specificaIly disclaims responsibility for any such activity. Hospitals. Ten general hospitals were ran-
domly selected from institutions of comparable size reporting to the Commission. They were located in the Northeastern and North Central United States. Nine were private nonprofit corporations; one was a city-county unit. Their identities were not disclosed. Five hospitals were major teaching institutions (MTH); the other 5 were nonaffiliated and had no residency training program (NAH). Four NAHs served nonmetropolitan areas (~50,000 population); the remaining NAH and aIl MTHs were in metropolitan areas (~ 100,000 population). Surgery, obstetrics, and newborn of MTHs admitted more patients than did the same services of NAHs (p ~ 0.05, rank-surn), but admission rates were similar in adult and pediatric medicine. Drug classes. Fifteen drug classes, as defined by the American Hospital Formulary System," were examined (Table I). Vasopressors were sympathomimetic agents, excluding ergot principles. Miscellaneous anti-infectives comprised urinary antiseptics, antitubercular and antiprotozoal agents, and anthelminthics. Vasodilators were products affecting the coronary and peripheral circulation. Tranquilizers included sedative and antipsychotic medications, except for single doses given preopera-
Hospital prescribing pattern
Volume 18 Number 2
135
Table 11. Mean prescribing rates 01 the major drug classes that constituted 96% 01 the pharmacotherapy surveyed in the pediatric, obstetric, and newborn services 01 JO general hospitals Drug utilization in per cent 01 alt (N) treatments
Patient exposure in per cent 01 alt (N) patients Drug classes Insulin Antihypertensi ves Diuretics Sulfonamides Vasopressors Tranquilizers Steroids Antibiotics Parenteral water and electrolytes
Pediatrics (N = 9965)
1.11 0.27 0.72 0.89 1.76 5.07 3.41 42.6 27.5
I
Obstetrics (N = 39,666)
I
Newborn (N = 21,275)
Pediatrics (N = 8456)
Obstetrics (N = 44,018)
Newborn (N = 1281)
0.01 0.21 0.33 0.01 1.80 0.02 0.05 2.22 1.24
1.31 0.32* 0.85* 1.05* 2.07 5.97 4.02 50.2 32.6
0.44* 0.66* 1.62 2.58 1.61 14.5 17.5 14.3 44.5
0.08* 3.51 5.54 0.16* 29.8 0.39* 0.78* 36.9 20.5
0.49 0.73 1.79 2.86 1.79 16.1 19.4 15.9 49.4
"This c1ass of drugs was not one of those that constitutcd 96% of the pharmacothcrapy prescribed in this spccialty.
tively or against nausea; they did not include barbiturates and chloral. Parenteral water and electrolytes referred to infusions of at least 1,000 ml per day of any fluid and solute, except blood, plasma, plasma protein fraction, and normal human serum albumin; the volume limit did not apply for children and infants. Diagnostic agents and a variety of special drugs (isotopes, dyes, radiopaques, local and general anesthetics, curare, anticonvulsants, antiparkinsonians, antidepressants, oxytocics, uricosurics, hematinics, antineoplastics, and immunosuppressants) were not included in this survey. Nor were nonspecific and symptomatic medications: analgesics, hypnotics, antihistamines, vitarnins, antitussives, antacids, antidiarrheals, laxatives, and antiemetics. Medications were inventoried by pharmacological classes instead of individual preparations. Thus, when a person received penicillin and streptomycin, this counted as 1 antibiotic treatment. Drug classes were divided into three categories according to the frequency of their use: a primary class represented more than 10% of all orders in a specialty; a secondary class, 1.3% to 6% of all prescriptions; and a tertiary class, less than 1.3% of the orders. Statistics. 1 MultisampIes were compared through Kruskal-Wallis' test if independent, or through Friedman's test if matched. Independent two-samples were compared through
Wilcoxon's rank-sum test, and paired observations through the matched-pair sign test. Correlations were estimated with Kendall's test; its index, T, was reported only when the probability of association was higher than 99%. Significance levels were determined for twosided probabilities when appropriate. Results
Hospitals. Average stay (p < 0.001, Friedman), mortality (p < 0.025), and mean drug exposure (p < 0.05) varied independently among the 10 institutions. These variations were not correlated, and together the 3 indicators did not reveal significant differences in overall performance among hospitals. The same result was obtained when the indicators were examined by specialties, except in medicine (p < 0.025, Friedman) where stay was related to mean drug exposure and mortality (p < 0.05, Kendall). No difference, however, was observed between teaching and nonteaching medical services. Total drug use. Mean drug exposure varied over a narrow range among medical (1. 34 to 2.94), surgical (1.04 to 1.90), and pediatric (0.68 to 1.12) services. The range in obstetrics (0.26 to 1.46) and newbom (0.02 to 0.13) was about 3 times as wide as in the other specialties. It should be noted that these exposures were relatively low because they omitted a number
136
Sice
Clinical Pharmacolog y and Therapeuties
of drug classes, especially analgesics and hypnotics." They were artifically reduced, also, by being inventoried according to pharmacological classes rather than individual medications. This data processing did not seem to affect comparative exposure to drugs, however, because 7 classes surveyed in these 10 medical services ranked exactly as in the medical service of another hospital where medications were also individually monitored." Such similarity between the results of collective and single drug inventories could be due to most patients having received several analogous medications in these 7 classes. Primary drug classes. Five classes of drugs accounted for 74% of the pharmacotherapy in medicine, and 3 of these for 75% of the orders in surgery (Table I). The other specialties relied even more heavily on primary drugs. Four classes accounted for 91 % of the orders in obstetrics, 2 for 83% of those in pediatrics, and 3 for 87% of those in newborn (Table II). Thus, most of the pharmacotherapy surveyed in these hospitals was represented by 7 classes of drugs, 3 of which were common to several specialties. Parenteral water and electrolytes were the most used drugs in surgery and obstetrics; they ranked second in medicine and pediatrics, and third in newborn. Antibiotics ranked first in pediatrics and newborn, second in surgery, and third in medicine. Medical patients often were given fluids with antibiotics (7 = 0.78), which suggests that in this specialty infusions might have been used principally as vehicles for the other drugs. Rates of fluids and antibiotics were uniform throughout the 10 hospitals. Tranquilizers were the most prescribed drugs in medicine, and more often used in this specialty than in any other. They ranked third in surgery and obstetrics. Their rate varied among hospitals (p < 0.05, Friedman), and was higher in medical and pediatric services of MTHs 0.05, rank-sum). than in those of NAHs (p Vasopressors were the most used drugs in 7 newborn services. Percentages of medical patients given tranquilizers (24.2 to 58.4), parenteral fluids (15.3 to 46.7), antibiotics (17.8 to 35.4), antiarrhyth-
=
mies (13.9 to 35.4), and diuretics (10.9 to 34.9) varied within fairly narrow ranges. People who took antiarrhythmics also received more medications (7 = 0.78) and stayed longer (7 = 0.78), but did not suffer a higher mortality than those who did not take cardiac agents. Percentages of surgical patients given tranquilizers (13.0 to 48.4), fluids (18.1 to 69.6), and antibiotics (26.8 to 38.4) varied within about the same ranges as in medicine. In pediatrics, the range of fluid use (8.3 to 44.1) was three times that of antibiotics (30.9 to 62.4). Primary drug use varied most in the remaining 2 specialties. In obstetrics, ranges of fluids (4.6 to 81. 3) and steroids (0.8 to 43.7) were several times as wide as those of tranquilizers (7.0 to 36.7) and antibiotics (8.0 to 23.5). In newborn, antibiotic (0.3 to 5.1), fluid (0.1 to 3.6), and vasopressor (0 to 4.6) use showed extreme variations among services. Secondary drug classes. Six classes of drugs accounted for 22 % of the pharmacotherapy in medicine, and 8 for 21 % of the orders in surgery (Table I). Secondary drug use was light in the other specialties: 3 classes accounted for 6% of the orders in obstetrics, 4 for 13% of those in pediatrics, and 2 for 9% of those in newborn (Table II). Four of these classes were common to several specialties. Diuretic and insulin use was uniform throughout the 10 hospitals-but steroid use varied widely among hospitals (p < 0.005, Friedman), though not between MTHs and NAHs. Vasopressor use also showed large variations among hospitals (p < 0.005, Kruskal-Wallis), and more patients received vasopressors in NAHs than in MTHs (p == 0.05, rank-sum). Percentages of patients given insulin (3.3 to 8.8) or antihypertensives (3.2 to 10.9) in medieine, insulin (I. 3 to 3.2) or antiarrhythmics (2.7 to 7.7) in surgery, tranquilizers (2.6 to 7.5) in pediatrics, and diuretics (0.8 to 2.6) or sulfonamides (1.0 to 4.8) in obstetrics remained within fairly narrow ranges. Other drug classes showed considerable variations. These occurred in medicine with steroids (1.7 to 12.0), anticoagulants (4.2 to 30.7), and vasodilators (2.0 to 16.9). In surgery, steroids (1.0 to 8.3), anticoagulants (1.0 to 7.7), di-
Volume /8 Number 2
uretics (2.3 to 12.3), and especially vasopressors (0.7 to 9.1), varied over wide ranges. Large variations also were observed in pediatrics with steroids (0.4 to 14.0) and vasopressors (0.4 to 5.6), in obstetrics with vasopressors (0.5 to 4.3), and in newborn with diuretics (0 to 1.2). Tertiary drug classes. Four cIasses of drugs in medicine and surgery each, 9 in pediatrics, and 8 in obstetrics and newborn each accounted for about 4% of the pharmacotherapy surveyed in every one of the specialties. The only significant difference observed in that category occurred in obstetrics, where antiarrhythmics and anticoagulants were prescribed more often in MTHs than in NAHs (p < 0.05, rank-sum). These differences, however, involved only about 0.5% of the patients with either class of drugs. Discussion
The objective of this survey was to provide a perspective of major drug-prescribing practices in large general hospitals. Two commonly used cIasses of drugs, analgesics and hypnotics," could not, however, be included in in the study because the data did not identify whether these medications had been given on or without formal doctors' orders. The results revealed more analogies than divergences in drug utilization, as might be expected considering the apparent homogeneity of the overall care given throughout these institutions, despite some variations among medical services and the wide ranges of mean drug exposure in obstetrics and newborn. In particular, no significant difference could be detected between teaching and nonteaching hospitals, except in their use of tranquilizers and vasopressors. The analogies included most of the drugs prescribed, beginning with parenteral fluids and antibiotics, which outclassed all treatments in surgery, pediatrics, and obstetrics, and yielded only to tranquilizers in medicine and to vasoconstrictors in newborn. The same consistency was noted with the 3 leading secondary classes of drugs: diuretics, antiarrhythmics, and insulin. Hence a consensus clearly prevailed among the hospitals regarding the administration of
Hospital prescribing pattern
137
61 % of their major pharmacotherapy. Such general agreement, however, does not necessarily prove that the use of these drugs always was valid." Agreement at the hospital level sometimes hid specific variations within certain specialties. Thus, prescriptions of fluids in pediatrics and obstetrics, of diuretics in surgery, and of antibiotics, fluids, and diuretics in newborn were not so uniform as in the other specialties, which also used these drugs. Such deviations could have been due to particular prescribing habits of the doctors regarding these medications, or to highly atypical requirements of the patients in a number of services. Tranquilizers provided a prime example of such anomalous pattern of drug use. They were prescribed almost twice as often by physicians as by surgeons, although patients admitted for an operation generally give the impression of being more anxious than those who are treated for a medical disorder. Then they were prescribed more often in medical and pediatric services of MTHs than in those of NAHs. This difference was not related to stay or to exposure to other drugs, and could not, therefore, be attributed to some exceptional difficulty of the diseases treated in teaching hospitals. Other explanations must therefore be sought in order to account for these prescriptions. It is possible, for instance, that patients were excessively distressed by the environment of MTHs, or these hospitals admitted people who were more anxious than those who went to NAHs. Practitioners in NAHs also might have handled psychoneurotic complaints with other means than sedatives, or they might have tolerated it with greater equanimity than did residents and teaching staffs. The data of this survey, however, do not allow testing of these hypotheses. Vasopressor use also deviated from normal prescribing patterns. First, it showed considerable inconsistency. In one institution, for instance, obstetrics and newborn were the lowest users of these drugs in these specialties, with 0.5% ofthe women and no infant receiving one of these agents; yet, surgery was the ranking prescriber of vasopressors in this specialty with 9% of its patients getting these medications. Such intramural differences could not
138
Clinical Pharmacology and Therape utics
Sice
easily be attributed to particular characteristics or needs of the population served by the hospital. Second, administration of vasopressors was unrelated to that of parenteral fluids. Third, vasopressors were prescribed more often in NAHs than in MTHs, although this practice did not correspond to apparently special requirements of patients in nonteaching institutions. Similarly erratic variations were observed with steroids, especially in pediatrics and obstetrics where these drugs were prescribed 35 times as often in some services as in others within the same specialty. Prescription of anticoagulants and vasodilators also showed a great inconsistency among medical services. But it was in newborn that rates of pharmacotherapy were least stable, and that these variations involved all drug classes, with coefficients ranging from 12 with diuretics and antihypertensives to 46 with vasopressors. Most of the drug classes that showed such atypical patterns of use were those for which therapeutic indications still are unclear or controversial. These findings indicate an urgent need for conducting critical investigations of the use, and adverse effects, of these drug classes in the major specialties, as is being done in medical services.v 6 Moreover, difficult or controversial pharmacotherapy should be evaluated by means of in-house controlled trials that would take into account the effects of local factors.
References 1. Bradley, J. V.: Distribution-free statistical tests, Englewood Cliffs, N. J., Prentice-Hall, Inc. 2. Constantino, N. V., and Kabat, H. F.: Druginduced modifications of laboratory test values, Am. J. Hosp. Pharm. 30:24-71, 1973. 3. Ellis, R. F., and Sice , J.:Availability and cost of urinary antibacterials in a metropolitan area, J. Chron. Dis. 26:617-622, 1973. 4. Jick, H.: Drugs-remarkably nontoxic, N. Eng!. J. Med. 291:824-828, 1974. 5. Maronde, R. F., Lee, P. V., McCarron, M. M., and Seibert, S.:A study of prescribing patterns, Med. Care 9:383-395, 1971. 6. May, F. E., Stewart, R. B., and Cluff, L. E.: Drug use in the hospital: Evaluation of determinants, CUN. PHARMACOL. THER. 16:834845, 1974. 7. Mazzullo, J.: The nonpharmacologic basis of therapeutics, CUN. PHARMACOL. THER. 13: 157158, 1972. 8. Sice, J.: Objectives of pharmacological education. In press, 1. Med. Educ. 9. Simmons, H. E., and Stolley, P. D.: Trends and consequences of antibiotic use in the United Stares, J. A. M. A. 227: 1023-1028, 1974. 10. Stolley, P. D., Becker, M. H., McEvilla, J. D., Lasagna, L., Gainor, M., and Sloane , L. M.: Drug prescribing and use in an American community, Ann. Intern. Med. 76:537-540, 1972. 11. Thur, M. P., Miller, W. A., and Latiolais, C. J.: Medication errors in a nurse-controlled parenteral admixture program, Am. J. Hosp. Pharm. 29:298-304, 1972.
THERAPEUTICS volume 18 number 2
August, 1975
Commentary Major drug-prescribing patterns in general hospitals Prescriptions of major drug classes were surveyed for one year in five teaching and five nonteaching general hospitals . Their overall rates varied over a two- to three-fold range in medicine, surgery, and pediatrics, but were uniform throughout the hospitals. Parenteral fiuids and antibiotics represented over one half of the pharmacotherapy used in every specialty except medicine, where tranquilizers were the mostfrequently prescribed drugs. Prescribing patterns lacked consistency in obstetrics and newborn services, and generally with steroids, vasopressors, and anticoagulants. Significant variations also involved the use offluids in pediatrics, diuretics in surgery, and vasodilators in medicine, These variations were too unpredictable and large to be attributed to specific differences among patient populations. Only two discrepancies were noted between teaching and nonteaching hospitals: tranquilizers were prescribed more often in the former, and vasopressors in the latter.
Jean Sice Cambridge, England Special Commission on Internal Pollution
Drug use in hospitals," clinics," and offices'? has received a great deal of attention during the last decade. These studies were conducted for a variety of reasons, the most urgent of which remain the dangers of medication abuse. Thus, the frequency of mishaps that complicate or prolong an illness has been shown to depend on the number of drugs taken." Complex regimens produce drug interactions," increase interferences with diagnostic tests," and multiply errors of dispensing!! and compliance. 7 Reprint requests 10: Dr. Jean Sice , 5139 Fifth Ave .. Pittsburgh, Pa. 15232.
The severity of adverse reactions, on the other hand, largely depends on the nature of the treatments: parenteral water and electrolytes, antiarrhythmics, tranquilizers, and anticoagulants account for the majority of lifethreatening accidents. 4 Antibiotics, also, can alter the characteristics of infectious diseases. 9 Other considerations involve the time spent in processing drug orders, starting and adjusting intravenous infusions, and the cost of pharmaceuticals." Knowledge of patterns of drug use, moreover, might help phannacology instructors to balance the importance given to the 133
134
Sice
Table I. Mean prescribing rates services 0/ 10 general hospitals
Clinical Pharmacolog y and The rape utics
0/ major drug
classes surveyed in medical and surgical
Patient exposure in per cent 0/ all (N) patients Drug classes
Thyroid and antithyroid agents Oral antidiabetics Vasopressors Miscellaneous anti-infectives Sulfonamides Insulin Antihypertensivcs Vasodilators Steroids Anticoagulants Diuretics Antiarrhythmies Antibiotics Tranquilizers Parenteral water and electrolytes
Medicine = 55,702)
(N
2.08 3.74 1.95 2.19 1.95 5.75 6.70 7.35 7.83 11.3 20.4 23.0 25.5 44.1 31.4
I
(N
Surgery = 87,782)
0.82 1.57 2.02 2.12 3.70 2.00 1.71 1.55 4.26 3.28 6.43 5.26 33.6 27.8
44.4
Drug utilization in per cent 0/ all (N) treatments Medicine (N = 108,713)
1.07* 1.92 1.00* 1.12* 1.00* 2.94 3.43 3.76 4.01 5.77 10.4 11.8 13.0 22.6 16.1
I
Surgery (N = 123.336)
0.59* 1.12* 1.43 1.51 2.63 1.42 1.21* 1.10* 3.03 2.34 4.57 3.74 23.9 19.8 31.6
*This class of drugs was one uf all those that consritutcd 4'K of the agents prescribed in this specialty.
various classes of drugs in educational programs." Most of the current information on prescribing, however, is derived from studies of general medical services in university hospitals. The pharmacotherapy practiced in the other specialties and in nonteaching institutions still is conjectural. The present survey, therefore , was undertaken in order to obtain this information. Its first objective was to determine prescribing rates of major drug classes in medicine, surgery, pediatrics , obstetries , and newborn in general hospitals. The second was to see if the use of these drugs conformed to some pattern, and to find if there might be consistent differences between teaching and nonteaching institutions. Methods
The data of this survey covered the entire calendar year 1972. They were obtained from the Commission on Professional and Hospital Activities, Ann Arbor, Michigan. But aIl analyses, interpretations, and conclusions based on these data are solely those of the author; the Commission specificaIly disclaims responsibility for any such activity. Hospitals. Ten general hospitals were ran-
domly selected from institutions of comparable size reporting to the Commission. They were located in the Northeastern and North Central United States. Nine were private nonprofit corporations; one was a city-county unit. Their identities were not disclosed. Five hospitals were major teaching institutions (MTH); the other 5 were nonaffiliated and had no residency training program (NAH). Four NAHs served nonmetropolitan areas (~50,000 population); the remaining NAH and aIl MTHs were in metropolitan areas (~ 100,000 population). Surgery, obstetrics, and newborn of MTHs admitted more patients than did the same services of NAHs (p ~ 0.05, rank-surn), but admission rates were similar in adult and pediatric medicine. Drug classes. Fifteen drug classes, as defined by the American Hospital Formulary System," were examined (Table I). Vasopressors were sympathomimetic agents, excluding ergot principles. Miscellaneous anti-infectives comprised urinary antiseptics, antitubercular and antiprotozoal agents, and anthelminthics. Vasodilators were products affecting the coronary and peripheral circulation. Tranquilizers included sedative and antipsychotic medications, except for single doses given preopera-
Hospital prescribing pattern
Volume 18 Number 2
135
Table 11. Mean prescribing rates 01 the major drug classes that constituted 96% 01 the pharmacotherapy surveyed in the pediatric, obstetric, and newborn services 01 JO general hospitals Drug utilization in per cent 01 alt (N) treatments
Patient exposure in per cent 01 alt (N) patients Drug classes Insulin Antihypertensi ves Diuretics Sulfonamides Vasopressors Tranquilizers Steroids Antibiotics Parenteral water and electrolytes
Pediatrics (N = 9965)
1.11 0.27 0.72 0.89 1.76 5.07 3.41 42.6 27.5
I
Obstetrics (N = 39,666)
I
Newborn (N = 21,275)
Pediatrics (N = 8456)
Obstetrics (N = 44,018)
Newborn (N = 1281)
0.01 0.21 0.33 0.01 1.80 0.02 0.05 2.22 1.24
1.31 0.32* 0.85* 1.05* 2.07 5.97 4.02 50.2 32.6
0.44* 0.66* 1.62 2.58 1.61 14.5 17.5 14.3 44.5
0.08* 3.51 5.54 0.16* 29.8 0.39* 0.78* 36.9 20.5
0.49 0.73 1.79 2.86 1.79 16.1 19.4 15.9 49.4
"This c1ass of drugs was not one of those that constitutcd 96% of the pharmacothcrapy prescribed in this spccialty.
tively or against nausea; they did not include barbiturates and chloral. Parenteral water and electrolytes referred to infusions of at least 1,000 ml per day of any fluid and solute, except blood, plasma, plasma protein fraction, and normal human serum albumin; the volume limit did not apply for children and infants. Diagnostic agents and a variety of special drugs (isotopes, dyes, radiopaques, local and general anesthetics, curare, anticonvulsants, antiparkinsonians, antidepressants, oxytocics, uricosurics, hematinics, antineoplastics, and immunosuppressants) were not included in this survey. Nor were nonspecific and symptomatic medications: analgesics, hypnotics, antihistamines, vitarnins, antitussives, antacids, antidiarrheals, laxatives, and antiemetics. Medications were inventoried by pharmacological classes instead of individual preparations. Thus, when a person received penicillin and streptomycin, this counted as 1 antibiotic treatment. Drug classes were divided into three categories according to the frequency of their use: a primary class represented more than 10% of all orders in a specialty; a secondary class, 1.3% to 6% of all prescriptions; and a tertiary class, less than 1.3% of the orders. Statistics. 1 MultisampIes were compared through Kruskal-Wallis' test if independent, or through Friedman's test if matched. Independent two-samples were compared through
Wilcoxon's rank-sum test, and paired observations through the matched-pair sign test. Correlations were estimated with Kendall's test; its index, T, was reported only when the probability of association was higher than 99%. Significance levels were determined for twosided probabilities when appropriate. Results
Hospitals. Average stay (p < 0.001, Friedman), mortality (p < 0.025), and mean drug exposure (p < 0.05) varied independently among the 10 institutions. These variations were not correlated, and together the 3 indicators did not reveal significant differences in overall performance among hospitals. The same result was obtained when the indicators were examined by specialties, except in medicine (p < 0.025, Friedman) where stay was related to mean drug exposure and mortality (p < 0.05, Kendall). No difference, however, was observed between teaching and nonteaching medical services. Total drug use. Mean drug exposure varied over a narrow range among medical (1. 34 to 2.94), surgical (1.04 to 1.90), and pediatric (0.68 to 1.12) services. The range in obstetrics (0.26 to 1.46) and newbom (0.02 to 0.13) was about 3 times as wide as in the other specialties. It should be noted that these exposures were relatively low because they omitted a number
136
Sice
Clinical Pharmacolog y and Therapeuties
of drug classes, especially analgesics and hypnotics." They were artifically reduced, also, by being inventoried according to pharmacological classes rather than individual medications. This data processing did not seem to affect comparative exposure to drugs, however, because 7 classes surveyed in these 10 medical services ranked exactly as in the medical service of another hospital where medications were also individually monitored." Such similarity between the results of collective and single drug inventories could be due to most patients having received several analogous medications in these 7 classes. Primary drug classes. Five classes of drugs accounted for 74% of the pharmacotherapy in medicine, and 3 of these for 75% of the orders in surgery (Table I). The other specialties relied even more heavily on primary drugs. Four classes accounted for 91 % of the orders in obstetrics, 2 for 83% of those in pediatrics, and 3 for 87% of those in newborn (Table II). Thus, most of the pharmacotherapy surveyed in these hospitals was represented by 7 classes of drugs, 3 of which were common to several specialties. Parenteral water and electrolytes were the most used drugs in surgery and obstetrics; they ranked second in medicine and pediatrics, and third in newborn. Antibiotics ranked first in pediatrics and newborn, second in surgery, and third in medicine. Medical patients often were given fluids with antibiotics (7 = 0.78), which suggests that in this specialty infusions might have been used principally as vehicles for the other drugs. Rates of fluids and antibiotics were uniform throughout the 10 hospitals. Tranquilizers were the most prescribed drugs in medicine, and more often used in this specialty than in any other. They ranked third in surgery and obstetrics. Their rate varied among hospitals (p < 0.05, Friedman), and was higher in medical and pediatric services of MTHs 0.05, rank-sum). than in those of NAHs (p Vasopressors were the most used drugs in 7 newborn services. Percentages of medical patients given tranquilizers (24.2 to 58.4), parenteral fluids (15.3 to 46.7), antibiotics (17.8 to 35.4), antiarrhyth-
=
mies (13.9 to 35.4), and diuretics (10.9 to 34.9) varied within fairly narrow ranges. People who took antiarrhythmics also received more medications (7 = 0.78) and stayed longer (7 = 0.78), but did not suffer a higher mortality than those who did not take cardiac agents. Percentages of surgical patients given tranquilizers (13.0 to 48.4), fluids (18.1 to 69.6), and antibiotics (26.8 to 38.4) varied within about the same ranges as in medicine. In pediatrics, the range of fluid use (8.3 to 44.1) was three times that of antibiotics (30.9 to 62.4). Primary drug use varied most in the remaining 2 specialties. In obstetrics, ranges of fluids (4.6 to 81. 3) and steroids (0.8 to 43.7) were several times as wide as those of tranquilizers (7.0 to 36.7) and antibiotics (8.0 to 23.5). In newborn, antibiotic (0.3 to 5.1), fluid (0.1 to 3.6), and vasopressor (0 to 4.6) use showed extreme variations among services. Secondary drug classes. Six classes of drugs accounted for 22 % of the pharmacotherapy in medicine, and 8 for 21 % of the orders in surgery (Table I). Secondary drug use was light in the other specialties: 3 classes accounted for 6% of the orders in obstetrics, 4 for 13% of those in pediatrics, and 2 for 9% of those in newborn (Table II). Four of these classes were common to several specialties. Diuretic and insulin use was uniform throughout the 10 hospitals-but steroid use varied widely among hospitals (p < 0.005, Friedman), though not between MTHs and NAHs. Vasopressor use also showed large variations among hospitals (p < 0.005, Kruskal-Wallis), and more patients received vasopressors in NAHs than in MTHs (p == 0.05, rank-sum). Percentages of patients given insulin (3.3 to 8.8) or antihypertensives (3.2 to 10.9) in medieine, insulin (I. 3 to 3.2) or antiarrhythmics (2.7 to 7.7) in surgery, tranquilizers (2.6 to 7.5) in pediatrics, and diuretics (0.8 to 2.6) or sulfonamides (1.0 to 4.8) in obstetrics remained within fairly narrow ranges. Other drug classes showed considerable variations. These occurred in medicine with steroids (1.7 to 12.0), anticoagulants (4.2 to 30.7), and vasodilators (2.0 to 16.9). In surgery, steroids (1.0 to 8.3), anticoagulants (1.0 to 7.7), di-
Volume /8 Number 2
uretics (2.3 to 12.3), and especially vasopressors (0.7 to 9.1), varied over wide ranges. Large variations also were observed in pediatrics with steroids (0.4 to 14.0) and vasopressors (0.4 to 5.6), in obstetrics with vasopressors (0.5 to 4.3), and in newborn with diuretics (0 to 1.2). Tertiary drug classes. Four cIasses of drugs in medicine and surgery each, 9 in pediatrics, and 8 in obstetrics and newborn each accounted for about 4% of the pharmacotherapy surveyed in every one of the specialties. The only significant difference observed in that category occurred in obstetrics, where antiarrhythmics and anticoagulants were prescribed more often in MTHs than in NAHs (p < 0.05, rank-sum). These differences, however, involved only about 0.5% of the patients with either class of drugs. Discussion
The objective of this survey was to provide a perspective of major drug-prescribing practices in large general hospitals. Two commonly used cIasses of drugs, analgesics and hypnotics," could not, however, be included in in the study because the data did not identify whether these medications had been given on or without formal doctors' orders. The results revealed more analogies than divergences in drug utilization, as might be expected considering the apparent homogeneity of the overall care given throughout these institutions, despite some variations among medical services and the wide ranges of mean drug exposure in obstetrics and newborn. In particular, no significant difference could be detected between teaching and nonteaching hospitals, except in their use of tranquilizers and vasopressors. The analogies included most of the drugs prescribed, beginning with parenteral fluids and antibiotics, which outclassed all treatments in surgery, pediatrics, and obstetrics, and yielded only to tranquilizers in medicine and to vasoconstrictors in newborn. The same consistency was noted with the 3 leading secondary classes of drugs: diuretics, antiarrhythmics, and insulin. Hence a consensus clearly prevailed among the hospitals regarding the administration of
Hospital prescribing pattern
137
61 % of their major pharmacotherapy. Such general agreement, however, does not necessarily prove that the use of these drugs always was valid." Agreement at the hospital level sometimes hid specific variations within certain specialties. Thus, prescriptions of fluids in pediatrics and obstetrics, of diuretics in surgery, and of antibiotics, fluids, and diuretics in newborn were not so uniform as in the other specialties, which also used these drugs. Such deviations could have been due to particular prescribing habits of the doctors regarding these medications, or to highly atypical requirements of the patients in a number of services. Tranquilizers provided a prime example of such anomalous pattern of drug use. They were prescribed almost twice as often by physicians as by surgeons, although patients admitted for an operation generally give the impression of being more anxious than those who are treated for a medical disorder. Then they were prescribed more often in medical and pediatric services of MTHs than in those of NAHs. This difference was not related to stay or to exposure to other drugs, and could not, therefore, be attributed to some exceptional difficulty of the diseases treated in teaching hospitals. Other explanations must therefore be sought in order to account for these prescriptions. It is possible, for instance, that patients were excessively distressed by the environment of MTHs, or these hospitals admitted people who were more anxious than those who went to NAHs. Practitioners in NAHs also might have handled psychoneurotic complaints with other means than sedatives, or they might have tolerated it with greater equanimity than did residents and teaching staffs. The data of this survey, however, do not allow testing of these hypotheses. Vasopressor use also deviated from normal prescribing patterns. First, it showed considerable inconsistency. In one institution, for instance, obstetrics and newborn were the lowest users of these drugs in these specialties, with 0.5% ofthe women and no infant receiving one of these agents; yet, surgery was the ranking prescriber of vasopressors in this specialty with 9% of its patients getting these medications. Such intramural differences could not
138
Clinical Pharmacology and Therape utics
Sice
easily be attributed to particular characteristics or needs of the population served by the hospital. Second, administration of vasopressors was unrelated to that of parenteral fluids. Third, vasopressors were prescribed more often in NAHs than in MTHs, although this practice did not correspond to apparently special requirements of patients in nonteaching institutions. Similarly erratic variations were observed with steroids, especially in pediatrics and obstetrics where these drugs were prescribed 35 times as often in some services as in others within the same specialty. Prescription of anticoagulants and vasodilators also showed a great inconsistency among medical services. But it was in newborn that rates of pharmacotherapy were least stable, and that these variations involved all drug classes, with coefficients ranging from 12 with diuretics and antihypertensives to 46 with vasopressors. Most of the drug classes that showed such atypical patterns of use were those for which therapeutic indications still are unclear or controversial. These findings indicate an urgent need for conducting critical investigations of the use, and adverse effects, of these drug classes in the major specialties, as is being done in medical services.v 6 Moreover, difficult or controversial pharmacotherapy should be evaluated by means of in-house controlled trials that would take into account the effects of local factors.
References 1. Bradley, J. V.: Distribution-free statistical tests, Englewood Cliffs, N. J., Prentice-Hall, Inc. 2. Constantino, N. V., and Kabat, H. F.: Druginduced modifications of laboratory test values, Am. J. Hosp. Pharm. 30:24-71, 1973. 3. Ellis, R. F., and Sice , J.:Availability and cost of urinary antibacterials in a metropolitan area, J. Chron. Dis. 26:617-622, 1973. 4. Jick, H.: Drugs-remarkably nontoxic, N. Eng!. J. Med. 291:824-828, 1974. 5. Maronde, R. F., Lee, P. V., McCarron, M. M., and Seibert, S.:A study of prescribing patterns, Med. Care 9:383-395, 1971. 6. May, F. E., Stewart, R. B., and Cluff, L. E.: Drug use in the hospital: Evaluation of determinants, CUN. PHARMACOL. THER. 16:834845, 1974. 7. Mazzullo, J.: The nonpharmacologic basis of therapeutics, CUN. PHARMACOL. THER. 13: 157158, 1972. 8. Sice, J.: Objectives of pharmacological education. In press, 1. Med. Educ. 9. Simmons, H. E., and Stolley, P. D.: Trends and consequences of antibiotic use in the United Stares, J. A. M. A. 227: 1023-1028, 1974. 10. Stolley, P. D., Becker, M. H., McEvilla, J. D., Lasagna, L., Gainor, M., and Sloane , L. M.: Drug prescribing and use in an American community, Ann. Intern. Med. 76:537-540, 1972. 11. Thur, M. P., Miller, W. A., and Latiolais, C. J.: Medication errors in a nurse-controlled parenteral admixture program, Am. J. Hosp. Pharm. 29:298-304, 1972.
