Between 1985 and 1999, as managed

Transcription

1 National In Use Of Medications In Office- Based Practice, Drugs were discussed 59 percent more often in office visits in 1999 than in 1985, rising to a rate of 146 prescriptions per 100 visits. by Catharine W. Burt ABSTRACT: Increases in physician office visits involving the use or prescribing of a drug were observed between 1985 and 1999 using data from the National Ambulatory Medical Care Survey. The prescription rate increased from 109 to 146 prescriptions per 100 visits. Growth in drug mention rates for specific therapeutic classes varied by patients age. The rate of multiple prescriptions per visit rose 39 percent. Similar-size increases were observed after differences in patients age, number of comorbidities, source of payment, and physician specialty were controlled for. Between 1985 and 1999, as managed care spread nationwide, more practice guidelines were developed for medication treatment. Changes in drug coverage, drug research and development (R&D), faster Food and Drug Administration (FDA) approval, and increased drug marketing to physicians and consumers are likely contributors to increased use of medications in ambulatory care. In this paper I use data from the National Ambulatory Medical Care Survey (NAMCS) to track physicians use of medications during this period. I describe trends in medication use at or after office visits, including patients and physicians characteristics that are related to increased reliance on medications. Methods Data in this study are from NAMCS, a multistage probability-based sample survey of 2,500 4,500 office-based physicians conducted by the National Center for Statistics (NCHS). NAMCS has collected patient encounter data periodically since 1973 and annually since For this report I performed a secondary analysis of survey data from 1985, , , , and Where possible, two years of data were combined to make the estimates more reliable. A total of 288,941 patient encounter records were used in this analysis. Details on the survey may be found elsewhere. 1 Physicians were asked to complete an encounter form for a sample of approximately thirty patient visits in a randomly assigned one-week reporting period during the survey year. Sample weights were applied to make national estimates of the number and characteristics of office visits. The response rates ranged from 80 percent in 1985 to 63 percent in Because NAMCS estimates are based on a sample of physician encounters, they are subject to sampling error. Standard errors were computed using SUDAAN, which takes the complex sample into consideration. 2 State- Catharine Burt is chief of the Ambulatory Care Statistics Branch at the National Center for Statistics, U.S. Centers for Disease Control and Prevention, in Atlanta. 206 July/August Project HOPE The People-to-People Foundation, Inc. Downloaded from Affairs.org on April 18, Copyright Project HOPE The People-to-People Foundation, Inc. For personal use only. All rights reserved. Reuse permissions at Affairs.org.

2 ments about significance are based on a weighted least-squares regression of estimates since The average relative standard error for the estimates presented here is.09. A multiple logistic regression analysis was performed using SUDAAN on the 1985 and 1999 data to model multiple prescribing events at office visits. The patient record form is used to collect data on patients characteristics such as age, sex, and race and on visit characteristics such as diagnosis, services provided, and medications prescribed. Physicians record all new or continued medications ordered, supplied, or administered at the visit, including prescription and nonprescription preparations, immunizations, desensitizing agents, and anesthetics. The maximum number of medications collected during the survey history varied from five to eight. This analysis used the first five drug mentions. Therapeutic class was assigned to the drug mentions based on the 1995 edition of the National Drug Code Directory (NDC). 3 In this analysis drugs classified in multiple categories are tabulated under the first category only. A report describing the method and instruments used to collect and process drug information has been published elsewhere. 4 I present unadjusted rates here because no significant differences appeared between drug mention rates that were ageadjusted and those that were not. Results The annual number of visits to office-based physicians increased 19 percent, from million in 1985 to million in The number of drug mentions during this time period increased 59 percent, from 693 million to 1.1 billion, with a resulting 34 percent increase in the drug mention rate from 109 to 146 drugs per 100 visits. The increase in drug mention rates was found for all age groups and all physician specialties except general surgeons, cardiologists, and dermatologists (Exhibit 1). The largest increase was found for psychiatrists, whose drug mention rate increased from 82 to 178 drugs per 100 visits. The distribution of visits to these specialties varied over time, as did some of the more common chronic diagnoses that are treated with medication therapy. The percentage of visits for hyperplipidemia and related lipoid disorders (such as high cholesterol and high triglycerides) had the largest increase (Exhibit 2). The number of medications at the selected chronic disease visits shown in the exhibit also increased. Increases were found for all of the major therapeutic classes, with the notable exception of antimicrobial agents, which declined 12 percent. The penicillins and tetracycline led the decline, and there was an increase in quinolones broad-spectrum antibiotics used to treat gastrointestinal, urinary tract, and lower respiratory tract infections. Six therapeutic classes account for 80 percent of the increase in the overall drug mention rate: central nervous system drugs, hormones, respiratory drugs, pain relief drugs, metabolics/nutrients, and cardiovascular-renal drugs (Exhibit 3). At a more detailed level, nine classes account for 75 percent of the observed increase (Exhibit 4). Increases in antidepressants alone account for 13.5 percent of the observed increase during the study period. Other classes contributing the most include lipid-lowering drugs and drugs to treat hypertension and diabetes. For comparability in this analysis, current therapeutic classes were assigned to 1985 drug mentions, because the 1995 NDC classification has more four-digit classes than there were in Drug mention increases varied by patient s age and therapeutic class (Exhibit 5). While the drug mention rate increased for almost all therapeutic classes for seniors, the largest increase was found for hematologic agents (187 percent, from 1.9 to 5.5 mentions per 100 visits). The largest increase observed for patients ages was 109 percent for metabolics (from 4.8 to 10.0 mentions), which includes lipid-lowering drugs such as Lipitor. Neurologic agents also showed a large increase among persons in this age group (from 1.6 to 2.6 mentions) and among those ages (from 2.4 to 4.7 mentions). Central nervous system drugs including selective serotonin reuptake inhibitors (SSRIs) had the largest in- HEALTH AFFAIRS ~ Volume 21, Number Downloaded from Affairs.org on April 18, Copyright Project HOPE The People-to-People Foundation, Inc. For personal use only. All rights reserved. Reuse permissions at Affairs.org.

3 EXHIBIT 1 Estimates Of Physician Office Visits And Drug Mentions, Selected Years Percent change since 1985 Number of visits (thousands) Visit rate per 100 persons Mean number of diagnoses Drug mentions (millions) 636, , , , , % a , , a 59 a Drug mentions per 100 visits All ages Under a 29 a 19 a 31 a 28 a 27 a Physician specialty General/family practice Internal medicine Pediatrics General surgery OB/GYN Orthopedic surgery Cardiovascular diseases Dermatology Urology Psychiatry Neurology Ophthalmology Otolaryngology All other a 25 a 19 a a 37 a a 116 a 36 a 78 a 39 a 33 a Percent of drug visits that have multiple medications 45.3% 45.2% 54.7% 54.5% 58.1% 28 a SOURCE: National Ambulatory Medical Care Survey, various years. NOTES: Drug mention rate is number of drug mentions divided by number of visits multiplied by 100. Estimates for , , and are annual averages. a Trend is significant (p <.01). crease for patients ages (130 percent, from 3.8 to 8.8 mentions). The therapeutic class with the largest increase for children was also central nervous system drugs, which increased 327 percent (from 1.1 to 4.5 mentions). Among the most frequently mentioned drugs in this class for children s visits in 1999 was Ritalin for treating attention deficit hyperactivity disorder (ADHD). The percentage of physician office visits with at least one medication mentioned increased by only 8 percent, from 61.2 to 66.2 percent of visits. This was driven by an increase for patients ages Multiple drug prescribing accounted for the largest part of the increase (Exhibit 6). Overall, the percentage of visits with two or more drugs increased 39 percent. A similar increase was found for each age group studied. A multiple logistic regression analysis modeled factors influencing multiple prescriptions among drug visits (for visits with any medications prescribed): patient s age, expected source of payment, number of diagnoses, and physician s specialty. Number of diagnoses at the visit measures comorbidity, which is related to the number of drugs prescribed. 5 Visits by persons with multiple conditions are more likely to require multiple medications. The percentage of visits with three diagnoses was 19.4 in 1999, up from 8.2 percent in Thus, some of the increase in drug prescribing 208 July/August 2002 Downloaded from Affairs.org on April 18, Copyright Project HOPE The People-to-People Foundation, Inc. For personal use only. All rights reserved. Reuse permissions at Affairs.org.

4 EXHIBIT 2 Percentage Of Physician Office Visits And Drug Mention Rates For Selected Chronic Conditions, 1985 And Diabetes (250) Drug mention rate Hyperlipidemia (272) Chronic respiratory (472, 473, 477) COPD ( ) Hypertensive disease ( ) Depression and related psychoses ( , 298.0, 298.1, 300.4) Percent of office visits for selected conditions NOTES: Drug mention rate is the number of drug mentions per 100 visits. ICD-9-CM codes are in parentheses and could be mentioned in any of three diagnoses collected. COPD is chronic obstructive pulmonary disease. EXHIBIT 3 Drug Mention Rates By Therapeutic Class, 1985 And 1999, With Change Since 1985 Number of drugs per 100 visits Therapeutic class Difference Percent distribution of differences All % 34% Central nervous system Hormones Respiratory Pain relief Metabolic/nutrients Cardiovascular/renal Ophthalmics Gastrointestinal agents Skin/mucous membranes Immunologics Neurologic agents Hematologic agents Antimicrobial agents Other classes Percent change since HEALTH AFFAIRS ~ Volume 21, Number Downloaded from Affairs.org on April 18, Copyright Project HOPE The People-to-People Foundation, Inc. For personal use only. All rights reserved. Reuse permissions at Affairs.org.

5 EXHIBIT 4 Top Detailed Therapeutic Classes Accounting For Most Of The Increase In Overall Drug Mention Rates At Physician Office Visits, 1985 And Antidepressants Hyperlipidemia drugs ACE inhibitors Antihistamines Calcium channel blockers Acid/peptic disorders Blood glucose regulators Vaccines Estrogens Number of mentions per 100 visits 6 7 NOTE: ACE is angiotensin converting enzyme. would be related to more visits in 1999 by patients with multiple diagnoses. Physician specialty was included in the model because of the large variation in drug therapies used in the diseases in which physicians specialize. The overall model explains 13 percent of the variation in multiple prescribing at drug visits. The odds that a drug visit had multiple drug mentions in 1999 over 1985 was 1.43 after adjusting for the all the factors in the model (95 percent confidence interval, 1.28, 1.61) (Exhibit 7). Discussion The NAMCS data examined here show increased use of medications since 1985 for all age groups, most specialties, and most therapeutic classes. The magnitude of the increases for specific therapeutic classes varied by age group. While one assumption is that increased medication use is due to aging of the population, differences in the proportions of middleage and senior patients during this time period account for only about 20 percent of the observed increase in overall drug mention rates. In 1985 patients age forty-five and older represented 42.1 percent of physician office visits. 6 In 1999 they represented 52.1 percent. 7 Adjusting the 1999 drug mention rate to the 1985 age distribution of office visits still leaves a 28 percent increase in drug mention rates. Another potential explanation is that doctors may be treating patients who, in general, have more comorbidities than in the past. The NAMCS data show that the relative proportion of visits with three diagnoses more than doubled during the study period. About half of the observed increase in drug mention rates can be explained by this change in case-mix. The percentage of visits with multiple drug mentions increased from 45.3 percent in 1985 to 58.1 percent in 1999, an overall increase of 28 percent. The logistic regression model shows that after differences in the distribution of visits by patients age, number of diagnoses, source of payment, and physician specialty are adjusted for, physicians were 43 percent more likely to prescribe multiple drugs in 1999 than in Examining other factors in the model also shows interesting evidence leading to multiple prescribing. Visits with government payment sources were more likely to receive multiple drugs (Medicare, odds ratio = 1.24; Medicaid, odds ratio = 1.44). Since these odds ratios are adjusted for differences in patients age and number of diagnoses, this may indicate that these patients are more likely to have con- 210 July/August 2002 Downloaded from Affairs.org on April 18, Copyright Project HOPE The People-to-People Foundation, Inc. For personal use only. All rights reserved. Reuse permissions at Affairs.org.

6 EXHIBIT 5 Drug Mentions Per 100 Visits, By Therapeutic Class And Patients Age, 1999, With Significant Percentage Changes Since 1985 Patients age (years) Therapeutic class Antimicrobial agents Cardiovascular/renal (87%) 6.5 (18%) Central nervous system Gastrointestinal agents 4.5 (327%) (130%) (74%) (64%) 8.0 (45%) 11.5 (27%) 10.3 (62%) Hematologic agents Hormones 0.3 a 3.0 (126%) 1.6 a (37%) 2.1 (75%) 20.7 (81%) 5.5 (187%) 21.0 (79%) Immunologics Metabolic/nutrients 25.4 (69%) 2.8 (73%) 4.5 (80%) (109%) 2.4 (58%) 17.2 (96%) Neurologic agents Ophthalmics (94%) (102%) (60%) 13.6 Pain relief Respiratory 8.0 (94%) 23.3 (39%) 9.8 (50%) 15.1 (81%) 16.9 (54%) 14.4 (73%) 17.9 (21%) 14.9 (62%) 19.0 (15%) 11.5 (41%) Skin/mucous membranes 9.3 (46%) (15%) 8.4 (50%) 7.4 (59%) NOTE: A lack of a percentage change indicates that there was not a significant linear trend since 1985 at the.01 level. a Estimate does not meet standard of reliability. EXHIBIT 6 Rise In Percentage Of Physician Office Visits With At Least One Drug Mention And Visits With Multiple Mentions, By Patients Age, At least 1 drug 2ormoredrugs Under 15 years years years years 65 years and older Percent change, HEALTH AFFAIRS ~ Volume 21, Number Downloaded from Affairs.org on April 18, Copyright Project HOPE The People-to-People Foundation, Inc. For personal use only. All rights reserved. Reuse permissions at Affairs.org.

7 EXHIBIT 7 Results From Multiple Logistic Regression In Modeling Multiple Drug Use In 1985 And 1999 For Visits At Which At Least One Drug Was Mentioned Visit characteristic Odds ratio 95% CI Age (ref = years) Under a a 65+ a Number of diagnoses (ref = 1) None 2 a 3 a Source of payment (ref = private insurance) Self-pay HMO Medicaid a Medicare a Other Physician specialty (ref = general/family practice) Internal medicine a Pediatrics General surgery a OB/GYN a Orthopedic surgery a Cardiovascular disease a Dermatology Urology a Psychiatry Neurology a Ophthalmology a Otolaryngology Other specialties a , , , , , , , , , , , , , , , , , , , , , , , , , 1.57 Year (ref = 1985) 1999 a , 1.61 NOTES: HMO is health maintenance organization. OB/GYN is obstetrics/gynecology. a Significant contributor to modeling whether a drug visit will have multiple drugs mentioned (p <.05) ditions that are amenable to drug treatment. The remainder of the observed increase could be driven by several factors, including the development of newly formulated drugs to treat chronic conditions, increased health insurance and prescription drug coverage, increased focus by physicians and health plans on managing health care, and direct-toconsumer marketing of drugs. New drugs. FDA approvals have doubled since the early 1980s, from an average of about nineteen per year to thirty-eight in the late 1990s. 8 Also, the length of time for approval of new drugs declined from about thirty-three months in 1986 to twelve months in This trend coincides with increased R&D by pharmaceutical companies. There were 104 new molecular entities (NMEs) approved between 1997 and 1999, many of which are used to treat chronic conditions. They accounted for 3.7 percent of all drug mentions in For example, among lipid-lowering 212 July/August 2002 Downloaded from Affairs.org on April 18, Copyright Project HOPE The People-to-People Foundation, Inc. For personal use only. All rights reserved. Reuse permissions at Affairs.org.

8 drugs, by 1989 statins replaced fibrates as the most heavily used medication. 10 Shifts from older statins to atorvastatin (Lipitor) also are associated with increased prescription patterns. Lipitor was first marketed in March Other frequently prescribed new drugs include Norvasc (first marketed in 1992), Claritin (1993), Zoloft (1992), Paxil (1993), Zocor (1992), Prempro (1995), and Glucophage (1995). These drugs were in the top twenty, by number of dispensed prescriptions in About half of the drugs mentioned in office visits in 1999 (and one-third of the drug mentions) were not mentioned in One of the main contributors to increases in drug prescribing are attributable to increased use of antidepressants, of which there have been at least eight new drugs since Drug coverage and managed care. Payment and delivery system issues may also be converging to influence the prescribing and use of medications. Payment sources for physician office visits have undergone huge changes since The NAMCS data show that in 1985, 35 percent of visits were paid for by the patient, in contrast to only 5 percent in The relative share of visits paid for by Medicare rose 40 percent and by private insurance or health maintenance organization (HMO), almost 200 percent. The percentage of insured workers having drug coverage rose from 91 percent in 1988 to 99 percent in As managed care made inroads into Medicare during the early 1990s, a higher proportion of seniors chose a plan option that provided some prescription drug benefit. To be competitive, many plans increased their drug coverage. More than half of all Medicare recipients now have some kind of drug coverage from supplemental plans. 14 Studies have shown that the likelihood of having drug coverage is directly related to the number of prescriptions a person needs. 15 The increased probability of filling a prescription may be related to physicians choice of medication over other therapies. The data from this study indicate that the drug mention rate increased at the same rate for all payment sources. If the 1999 visits were distributed like the 1985 visits with regard to source of payment, the same level of increase in total drug mention rate would have been observed. The effect of managed care on drug prescribing may be more subtle. Increased emphasis on managed care has also likely led to increased use of treatment guidelines, many of which included medical consensus on medication use. The proliferation of practice guidelines developed and supported by third-party payers would help to explain the increased reliance on medication therapy. For example, treatment guidelines for hypertension include multiple drug regimens in many cases. 16 Similarly, there has been an increased emphasis on medications for pain relief. 17 While approximately one-quarter of office visits include mention of the patient s pain, the relative percentage of those visits that included mention of a pain medication increased 21 percent, from 24 percent of visits in 1985 to 29 percent in 1999 (data not shown). The percentage of visits with no direct pain complaint that had a mention of pain medication increased from 6 percent to 9 percent during the study period (data not shown). DTC advertising. Direct-to-consumer (DTC) advertising for prescription drugs blossomed in 1997 after the FDA clarified rules pertaining to ads, making it easier to launch television, print, and radio ad campaigns. Such increased mass-media advertising has coincided with a sharp rise in the number of prescriptions written and overall spending on prescription drugs. Doctors wrote 34 percent more prescriptions in 1999 than in 1998 for the twenty-five DTC-promoted drugs that contributed most to overall drug spending, while they wrote only 5 percent more prescriptions for all other prescription drugs. 18 For example, Pfizer increased DTC advertising for Lipitor sixfold between 1998 and 1999 and saw prescriptions increase by 50 percent. Of course, pharmaceutical companies may have chosen to apply DTC advertising to the new drugs that were expected to have the most use. Nevertheless, as DTC advertising has increased over the years, so has the number of medications prescribed increased, whether DTC advertising motivates patients to come to the doctor or HEALTH AFFAIRS ~ Volume 21, Number Downloaded from Affairs.org on April 18, Copyright Project HOPE The People-to-People Foundation, Inc. For personal use only. All rights reserved. Reuse permissions at Affairs.org.

9 motivates doctors to prescribe more drugs at the patient s request. 19 The use of medications in treating medical conditions is likely to continue to rise as the baby-boom generation stretches from middle to old age. Scientific breakthroughs will find applications in better-engineered drugs to treat many other conditions. Better treatment of acute and chronic conditions is associated with increased life expectancy. However, this will lead to even greater dependence on such drugs for maintaining a high quality of life. Data from the Chain Pharmacy Industry Profile show that in drug spending rose by 165 percent with the most recent average yearly increase about 15 percent. 20 At that rate, drug spending will likely double in the next five years, despite patent expirations for many popular brand-name drugs. NOTES 1. See the annual summaries that are part of the Advance Data from Vital and Statistics publications, National Center for Statistics (NCHS), adata.htm (18 April 2002). 2. B.V. Shah et al., SUDAAN User s Manual, Release 7.0 (Research Triangle Park, N.C.: Research Triangle Institute, 1996). 3. U.S. Food and Drug Administration, National Drug Code Directory, 1995 ed. (Washington: Public Service, 1995). 4. H. Koch, The Collection and Processing of Drug Information: National Ambulatory Medical Care Survey, United States, 1980, Vital and Statistics, Series 2, no. 90, DHHS Pub. no. (PHS) (Washington: U.S. Government Printing Office, March 1982). 5. In 1999 the drug mention rate at visits with three diagnoses was mentions per 100 visits; at visits with only one diagnosis, the rate was mentions. Up to three diagnoses are collected at each visit in NAMCS and coded according to the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) (Washington: PHS, 1999). 6. T. McLemore and J. DeLozier, 1985 Summary: National Ambulatory Medical Care Survey, Advance Data from Vital and Statistics, no. 128 (Hyattsville, Md.: National Center for Statistics, 2001). 7. D. Cherry, C. Burt, and D. Woodwell, National Ambulatory Medical Care Survey: 1999 Summary, Advance Data from Vital and Statistics, no. 322 (Hyattsville, Md.: NCHS, 2001). 8. Henry J. Kaiser Family Foundation, Prescription Drug : A Chartbook (Menlo Park, Calif.: Kaiser Family Foundation, July 2000). 9. Cherry et al., National Ambulatory Medical Care Survey. 10. T.J. Wang et al., Randomized Clinical Trials and Recent Patterns in the Use of Statins, American Heart Journal 141, no. 6 (2001): Kaiser Family Foundation, Prescription Drug. 12. T.W. Croghan, The Controversy of Increased Spending for Antidepressants, Affairs (Mar/Apr 2001): U.S. Department of and Human Services, Report to the President: Prescription Coverage, Spending, Utilization, and Prices (Washington: DHHS, April 2000). 14. A.S. Adams, S.B. Soumerai, and D. Ross-Degnan, The Case for Making a Medicare Drug Coverage Benefit, Annual Review of Public 22 (2001): J.A. Poisal and L. Murray, Growing Differences between Medicare Beneficiaries with and without Drug Coverage, Affairs (Mar/Apr 2001): C.R. Nelson and D.A. Knapp, in Antihypertensive Drug Therapy of Ambulatory Patients by U.S. Office-Based Physicians, Hypertension 36, no. 4 (2000): Agency for care Research and Quality, Acute Pain Management: Operative or Medical Procedures and Trauma Clinical Practice Guideline, Pub. no (Rockville, Md.: AHRQ, February 1992). 18. S. Findlay and D. Sherman, Prescription Drugs and Mass Media Advertising, NIHCM Foundation Research Brief (Washington: National Institute for Care Management, September 2000). 19. Cherry et al., National Ambulatory Medical Care Survey. 20. Kaiser, Prescription Drug. 214 July/August 2002 Downloaded from Affairs.org on April 18, Copyright Project HOPE The People-to-People Foundation, Inc. For personal use only. All rights reserved. Reuse permissions at Affairs.org.