Microsoft word - working papers--link-3-08.doc

ALFRED LERNER COLLEGE OF BUSINESS & ECONOMICS WHO RECEIVES STATINS? VARIATIONS IN PHYSICIANS’ PRESCRIBING PATTERNS FOR PATIENTS WITH CORONARY HEART DISEASE, DYSLIPIDEMIA, AND DIABETES Simon Condliffe, Charles R. Link and Bryan Townsend ____________________________*http://lerner.udel.edu/economics/workingpaper.htm. 2007 by author(s). All rights reserved.
WHO RECEIVES STATINS? VARIATIONS IN PHYSICIANS’ PRESCRIBING
PATTERNS FOR PATIENTS WITH CORONARY HEART DISEASE,
DYSLIPIDEMIA, AND DIABETES
*Simon Condliffe: Center for Applied Demography and Survey Research, University of
Delaware; Charles R. Link: Department of Economics, University of Delaware, 408
Purnell Hall, Newark, Delaware 19716 (e-mail: linkc@lerner.udel.edu; TEL: (302) 831- 1921; FAX: (302) 831-6968); Bryan Townsend, University of Delaware.
Our objective is to estimate the extent to which clinical and non-clinical factors areassociated with physicians’ prescribing patterns for statins. The data are from theNational Ambulatory Medical Care Survey for the period 1992 through 2004. The threesamples examined included more than 14,000 patients who were diagnosed with coronaryheart disease, high cholesterol, or diabetes, individuals who are most likely to benefitfrom being prescribed a statin drug. Using a multinomial logit framework, we finddisparities in prescribing patterns based on non-clinical factors. Namely, whites andpatients who have private insurance are more likely to be prescribed a statin than non-whites and those with public insurance. Also, even though a large increase occurred inthe uptake of statins over the period 1992 to 2004, our results for 2004 show that onlyabout 50 percent of patients diagnosed with coronary heart disease were prescribed astatin. Because coronary heart disease is the leading cause of death in the U.S. andcurrently is estimated to cost over $150 billion annually in the U.S. in direct and indirectcosts, observed differences in prescribing patterns along these dimensions is troublingand should be part of discussions dealing with health care reform.
Key Words: Pharmaceuticals; Statins; Equity in Physician Prescribing Patterns; Insurance The objective of the paper is to estimate the factors that influence whether physicians prescribe statins and/or other lipid-lowering drugs to patients diagnosed with coronary heart disease (CHD), dyslipidemia, and diabetes. In 2004, coronary heart disease (CHD) affected 15.8 million people in the U.S., and is the leading cause of deaths with over 450,000 deaths annually. Not only is CHD the leading cause of death, but in 2007 it is estimated to cost over $150 billion in direct and indirect costs (American Heart Association, 2007).1 There are several risk factors for CHD, of which diabetes and high levels of cholesterol, low-density lipoprotein cholesterol (LDL), are relevant to the Given the well-documented success of statins in lowering cholesterol levels, do people at risk for CHD have equal access to statins? That is, do disparities in prescribing patterns of physicians exist across such factors as the type of insurance the patient has, where the patient lives, or the patient’s ethnic/minority status. Doty and Holmgren (2006), based on the Commonwealth Fund Biennial Health Insurance Survey (2005), found that uninsured rates for Hispanic and African Americans are one and one-half to three times higher than the rate for whites. A recent report by the Institute of Medicine (IOM) (Smedley, Stith, and Nelson, 2002) has documented the inequalities in health care received by minorities in the U.S., even in cases where there is similar access to health care. Both the IOM study and the study by the Kaiser Foundation and the American College of Cardiology Foundation (2002) found that minorities are less likely than whites to receive invasive procedures such as angioplasty, bypass surgery, and thrombolytic 1 Direct costs include hospitals, nursing home, physicians/other professionals, and drugs, and other medicaldurables and home health care. Indirect costs include lost productivity/morbidity and lostproductivity/mortality.
therapy. The existence of racial disparities in care is important enough to have resulted in a whole issue of Health Affairs (2005) being devoted to the topic. Results of our study of factors associated with physician prescribing patterns for statins will shed light on the potential disparities in regards to the access to statins by Black Americans and Hispanics.
The importance of equity in prescribing practices is magnified when a drug is effective in treating a serious medical condition. Numerous studies have shown that statins are more effective than predecessors of similar function in decreasing both LDL cholesterol and triglycerides, which are important risk factors for heart disease. Recent studies have also suggested that statins may serve not only as a treatment for reducing high LDL cholesterol levels, but also in the prevention of other conditions such as stroke, deep vein thrombosis, dementias (including Alzheimer’s disease), and as treatment to slow the progression of multiple sclerosis (Noonan, 2003, Manning, 2004). We were unable to find other multivariate studies dealing with prescribing patterns by physicians as they relate to statins, so our paper provides initial evidence relating to the equity issues According to Pearson (2007), compliance studies indicate that patients often do not follow up on the therapies prescribed by their physicians. Since our study deals only with whether or not the physician prescribes a statin drug, and we do not know whether the patient actually filled the prescription, our results provide upper-bound estimates of the proportion of the patients who actually end up taking the drugs. Nonetheless, our findings should be of interest to policy makers whose objective is to reduce racial/ethnic disparities in cardiac care and health outcomes in general.
Section 1 includes a brief discussion of why the particular diagnoses were chosen.
Section 2 discusses the methodology, including the data source, variables included in the analysis, and econometric models underlying the estimations. Results are shown in section 3. Section 4 includes a summary and conclusions.
1. Why CHD, Dyslipidemia, and Diabetes?
Cardiovascular related diseases have been for many years and continue to be the leading cause of death for males and females in the United States (American Heart Association, 2007). In 2004, in the U.S., it was estimated that 79.3 million people age 20 and older had LDL cholesterol levels considered ‘borderline high’ or worse; and high LDL cholesterol levels are a risk factor for CHD (Centers for Disease Control, 2006a, 2006b, American Heart Association, 2007).2 The incidence of and death rates associated with cardiovascular diseases are greater for Blacks compared to whites (American Heart Diabetes affects over 20 million people in the United States and was the sixth leading cause of death in the U.S. in 2004. Type-2 diabetes is the most common type, with risk factors including obesity, inactivity, older age, high blood pressure, high cholesterol, a family history of diabetes, and ethnicity. Compared to Caucasians, Hispanic/Latino Americans, African Americans, and Asian Americans are twice as likely to develop Type-2 diabetes (1 on 1 Health, Centers for Disease Control, 2003, 2005, 2 Levels of 130-159 are borderline high.
Statins have been effective in lowering LDL cholesterol levels. Use of statins as sole means of primary or secondary treatment has been shown to reduce the incidence of CHD by 25 to 60 percent and reduce the risk of death from any other heart disease or condition by 30 percent (Knopp, 1999). The use of statins, in addition to providing large and significant reductions in deaths due to CHD, does not appear to increase non-CHD or cancer related deaths (Hebert et al, 1997). Reductions in risk for CHD were similar for men and women and for elderly and middle-aged patients (LaRosa et al, 1999). Although these results suggest the potential for reducing cholesterol levels in large numbers of persons at risk for CHD, other studies have shown that only one-third of eligible patients are receiving statins as secondary prevention (Majumdar et al, 1999). Data from the Third Report of the Expert Panel on Detection, Evaluation and Treatment of High Blood Cholesterol in Adults (2007) cites results consistent with Majumdar and further notes that even among the highest risk patients less than half are getting lipid lowering medications.
Specific at risk populations in danger of not receiving statins include minorities and patients lacking insurance coverage for prescription drugs.
One of the most detrimental side effects of diabetes is heart disease and stroke, the leading cause of death among diabetics. For those having diabetes, heart disease occurs earlier in life, is two to four times more common than in non-diabetics, and is more likely to be fatal. The use of statins to control cholesterol levels in diabetics is an increasingly important method of allowing patients to live with diabetes.3 3 Studies suggest that all diabetics, including those with high blood pressure, wouldbenefit from the use of a statin. One study included approximately 6,000 diabeticpatients and only half of these 6,000 showed signs of cardiovascular disease at the time ofthe study. Interestingly, it found that a daily dose of simvastatin (40 mg) decreased therisk of an initial heart attack or stroke by around 30% for diabetes patients and also 2. Data and Methodology
The data set utilized in the current study is from the National Ambulatory Medical Care Survey (NAMCS) for the years 1992 through 2004 (Schappert S.M. & Nelson C.R.
(1999); U.S Department of Health and Human Services (2000)). The NAMCS includes data from patient visits to non-federally employed office-based physicians who focus mainly on direct patient care.4 Data for selected patients are entered onto the patient record form and include diagnoses, treatments, symptoms, services performed, prescriptions given, and various pieces of demographic information. The physician or office staff, and not the patient, enter data into the survey forms. Specially trained interviewers meet with physicians prior to survey participation, and provide them with materials and instruction on how to complete the instrument. Each physician is randomly assigned to a one-week reporting period, during which data for a random sample of visits reduced the probability of a repeat attack in those who had already suffered one (TheDoctor's Lounge, 2003). Other studies of statins and diabetes yielded similar results. Dr.
Collins as a result of his study, specifically directed towards diabetics states, “Loweringyour cholesterol will lower your risk of heart disease, irrespective of what yourcholesterol level is. Taking a statin will lower your risk by a third. And if you continueto take it, you will continue to lower your risk.” (DeNoon, 2003) 4 Certain specialties, including anesthesiology, pathology, and radiology are not included in the survey.
Data was obtained on the National Center for Health Statistics (NCHS) website in the format of self-extracting files, which contained all survey data as well as descriptions of the survey, record format, codesfor medications, generic drugs, and various other information files.
5 See http://www.cdc.gov/nchs/about/major/ahcd/namcsdes.htm Three samples we examine include 6,205 patients whose primary diagnosis is coronary heart disease, 2,057 with dyslipidemia, and 6,215 diagnosed with diabetes. The primary diagnosis is based on the International Classification of Diseases (ICD-9) coding scheme. Patients with coronary heart disease were those whose primary diagnosis code fell between 410.00 and 414.90. These patients are broadly defined as having ischemic heart disease.6 Patients with dyslipidemia were those whose primary diagnosis code fell between 272.00 and 272.90, a group that is broadly defined as having disorders of lipid metabolism.7 The patients who had an ICD-9 code having the first three digits 250 include a group that is broadly defined as having type-1 or type-2 diabetes.8 A patient’s insurance coverage and source of payment for the office visit was indicated by one of several options. The insurance variables were recoded so that a patient fell into one of five insurance categories, reflecting the likelihood that the patient’s insurance provided prescription drug coverage. These groups are: 1) Medicaid, 2) HMO/Private, 3) Medicare, 4) self-pay, and 5) unknown/blank. In cases when a patient’s record indicated more than one type of insurance, the patient was placed in the group among those indicated on his or her record that was most likely to provide 6 More specific sub-groups include acute myocardial infarction (410), other acute and subacute forms of ischemic heart disease (411), old myocardial infarction (412), angina pectoris (413), and otherforms of chronic ischemic heart disease (414).
7 Specific sub-groups within this coding range include hypercholesterolemia (272.0), hypertriglyceridemia (272.1), mixed hyperlipidemia (272.2), hyperchylomicronemia (272.3), otherhyperlipidemia (272.4), lipoprotein deficiencies (272.5), lipodystrophy (272.6), lipidoses, (272.7), otherdisorders of lipid metabolism (272.8), and unspecified disorders of lipid metabolism (272.9).
8 Patients with gestational diabetes were not included in our analysis.
prescription drug coverage. For example, if the record indicated Medicaid and Medicare, the patient was placed in the Medicaid group.
The Logit Model
We estimate a multinomial logit model of the probability of a patient being prescribed a statin in which independent variables, described more fully below, include demographic characteristics, medical condition, and insurance status. The dependent variable has three outcomes: no treatment, prescribed a statin9, and prescribed a lipid- Demographic explanatory variables utilized in the logit regressions are dummy
variables for calendar year; patient’s age (<45, 45-64, and 65-74); gender; race/ethnicity (Black, Asian, and Hispanic); and residence (Midwest, South, and West; Nonmetro).
Insurance status is categorized into four categories: Medicaid, Medicare, Selfpay, and
Unknown. Classification of secondary diagnoses is based on the patient’s secondary
diagnosis in the ICD-9 group: CHD2 (codes 410.00 to 414.90); Hypertension (401.00
to 405.90); Other Heart Disease (420.00 to 429.90); Vascular Disease (440.00 to
448.90); Diabetes (250.00 to 250.93); Dyslipidemia (272.00 to 272.90); and All Other
ICD-9 groups. For the Diabetes sample only, we also include a variable to measure whether a patient’s secondary diagnosis is both dysliplidemia and coronary heart disease.
9 Statins include Atorvastatin or Lipitor, Cerivastatin or Baycol, Fluvastatin or Lescol,Lovastatin or Mevacor (Altocor), Pravastatin or Pravachol, Rosuvastatin or Crestor,Simvastatin or Zocor, Simvastatin + Ezetimibe or Vytorin. Note that Baycol waswithdrawn from the market in 2001. The FDA approved Crestor in August 2003 andVytorin in July 2004.
10 The following drugs are included in the other lipid lowering drugs: Cholestyramine,Questran, Lo-Cholest, Prevalite, Colesevelam,WelChol, Colestipol,Colestid, Fenofibrate,Tricor, Gemfibrozil, Lopid, Clofibrate, Atromid S, Niacin (nicotinic acid),Ezetimibe,Zetia, and Ezetrol.
The omitted groups in the logit regressions include the following variables: the year 1992, older than age 74, female, white, paid through private insurance (“HMO/Other prepaid,” “Private/Commercial Insurance,” “Blue Cross/Blue Shield,” or “Other private insurance”), northeast, metro area, and had no co-morbidities (secondary diagnoses).
Time dummy variables are included to give us information about the uptake and diffusion of statins over the period 1992 through 2004. We expect their coefficients will indicate increased use of statins over time. Age is a risk factor for CHD and diabetes.
Clinical guidelines indicate that persons in our three primary condition categories should be treated more aggressively for high cholesterol if they exhibit secondary conditions (risk factors) such as hypertension, dyslipidemia (for patients with CHD and diabetes), and CHD (for patients with diabetes and dyslipemia). Positive coefficients are expected for the variables indicating risk factors for CHD. The race/ethnicity variables indicate the presence or absence of disparities in access to statins along the lines found by the Institute of Medicine in regards to cardiac care for minorities. Although males are more likely than females to have CHD, and since statin treatment is effective in both genders, we have no a priori prediction for the sign of the coefficient. The variables indicating the method of payment for the office visit will shed light on the importance of the source of payment as a potential determinant of whether or not a patient is prescribed a statin drug by his or her physician. We expect that private insurance will be associated with a greater likelihood of the patient receiving a statin. Information about a person’s residence--the region of residence in the U.S. and whether he or she lives in a metropolitan area--will allow us to determine whether or not where one lives (which is not part of clinical guidelines) affects who is prescribed a statin.
3. Results
In this section we first show the demographic and clinical conditions associated with each of the three samples under study. The logit results are then discussed.
Sample Means
Table 1 shows the means for variables associated with the three samples of respondents diagnosed with a primary condition of coronary heart disease (CHD, column 1), dyslipidemia (DYS, column 2), and diabetes (column 3). Since males are more likely to have CHD, the fact that they comprise 63 percent of the sample diagnosed with CHD is not surprising. The percent male drops to around 50 percent for persons diagnosed with dyslipidemia or diabetes. Whites make up more than 83 percent of the CHD and DYS samples but only about 70 percent of diabetics. Blacks comprise between six to seven percent percentage of the CHD and DYS samples, while for Hispanics the percentages are between 4.7 and 5.8 percent. These percentages for diabetics jump to 10 percent for Hispanics and 15.9 percent for Blacks. The low percentages of Blacks and Hispanics in the coronary heart disease and dyslipidemia samples are consistent with evidence that minorities comprise a smaller percentage of the samples visiting physicians’ offices in ambulatory settings when compared to the percentage they comprise of the whole population in the U.S. This is probably due to Medicaid, where the number of solo physicians not accepting new Medicaid patients over the period 1996- 1997 to 2004-2005 has risen from 29 to 35.3 percent. Similarly for small groups of physicians, the percentage declining new Medicaid patients rose from 16.2 percent to 24 percent over the same period (Cunningham and May, 2006). Reasons for the increased likelihood of physicians’ refusal to take on Medicaid patients include the low payments compared to private insurance along with high administrative costs. More than half of the 52 million Medicaid beneficiaries are from minority groups, more than 20 percent of non-elderly African Americans and Latinos (Llanos and Palmer, 2006).
Method of payment for the physician visit varies by primary diagnosis. Medicaid patients comprise 5 percent of the visits for CHD, 2.5 percent for dyslipidemia, and 9.9 percent for diabetes. The percentages for Medicare are 45 percent for CHD, 32 percent for diabetics, and 25 percent for those diagnosed with dyslipidemia. Payment by private insurance varies from around 45-48 percent for CHD and diabetes, to 62 percent for dyslipidemia. The age composition of the sample with CHD rises from 3.5 percent for those under age 45, to around one-third for each of the three age groups 45-64, 65-74, and 75 and older. The distribution by age differs for patients with dyslipidemia and diabetes where the peak of 42 to 47 percent occurs in the 45 – 64 age-group.
Approximately one third of the patients in each diagnosis category resided in the South, while the remainder was evenly divided between the other three regions of the U.S. The vast majority of patients lived in metro areas.
Within each diagnosis, the presence of co-morbidities is likely to influence physicians’ decisions about whether or not to prescribe lipid-lowering drugs. Of persons diagnosed with CHD, 11 percent also had other heart conditions, 15.1 percent had dyslipidemia, 6.5 percent had diabetes, and 13.8 percent had hypertension. Patients with dyslipidemia and diabetes were most likely to have hypertension as the only other secondary condition. About 40 percent of the patients diagnosed with dyslipidemia and diabetes had a secondary condition unrelated to the conditions included in table 1. The percentages of respondents diagnosed with CHD, dyslipidemia, and diabetes, but who have no secondary conditions, are 27, 36, and 23 percent respectively.
Logit results
The results from the logit regressions are shown in table 2 columns 1-2 for CHD, columns 3-4 for dyslipidemia, and columns 5-6 for diabetes. Table 3 shows the resulting The coefficients for the year dummy variables should be interpreted with respect to 1992. As expected, statin utilization rates increased over time for each diagnosis category, and uptake in their usage reflects a substitution for other lipid-altering drugs and a decrease in the probability of receiving ‘no treatment.’ Evaluating the logit results at the mean values, the probability of receiving a statin in 1992 if the patient had CHD, dyslipidemia, or diabetes as the primary condition, was respectively 13.3 percent, 20.9 percent, and 3.3 percent. By 2004 these probabilities rose to 50.3 percent, 51.6 percent and 28 percent respectively (see Table 3). The probability of a patient diagnosed with CHD not receiving any treatment dropped from 82 to 48 percent.
For persons diagnosed with coronary heart disease, coefficients for all age groups are positive and statistically significant at the 10 percent level or higher (2 of 3 significant at the 5 percent level), indicating these patients are more likely to receive statins than are patients 75 and over. Although the coefficients are not all statistically significant, a similar pattern is observed for other lipid lowering drugs. This pattern is likely to reflect the controversy surrounding statins and their use on the elderly. While studies have shown benefits of reducing LDL cholesterol in reducing the risk of heart attacks and stroke, some research suggests that lowering cholesterol in the oldest age group may increase the risk of death (MayoClinic.com, 2005).
Persons with dyslipidemia under the age of 45 were less likely to receive a statin than were the oldest patients. Being in the 65 - 74 age cohort increases the probability of being prescribed a statin compared to people in the over 75 age group. There is no age pattern in the prescribing patterns for other lipid lowering drugs. In the case of diabetics, no significant differences by age existed in prescribing patterns for statins. Diabetics aged from 45 through 74 are more likely than the oldest cohorts to be prescribed other Since men are generally at higher risk of developing CHD than women in the same age group, we might expect that statins would be more likely to be prescribed to men (other factors the same).11 Although the coefficient is positive, it is only significant at the 10 % level, thus providing little support for this conclusion. For those with dyslipidemia and diabetes, gender is not associated with prescribing patterns of statins or Race/ethnicity
Blacks and Hispanics with CHD are less likely than whites to be prescribed statins, with the coefficients of each negative and significant at the 10% level. Hispanics and Asians with dyslipidemia are also less likely than whites to receive a statin, with 11 See http://www.mayoclinic.com/health/coronary-artery-disease/DS00064 coefficients significant at the 5% level. There are no statistically significant racial differences in prescribing patterns for statins for diabetes patients in the time period under study. The only racial/ethnic differences in the ‘other lipid lowering drugs’ occurred with Black and Hispanic persons diagnosed with diabetes, where they are less likely than whites to be prescribed OLAD drugs.
Table 3 presents the predicted probabilities of being prescribed a statin drug by race/ethnicity group in 2004 for those with CHD, the main diagnosis group where race disparities exist. The probabilities for Blacks, Hispanics, and whites in 2004 are respectively 45 percent, 43.3 percent, and 50.9 percent. For persons with dyslipidemia, the corresponding percentages are 53.8, 43.6, and 51. For diabetics, the figures for Blacks, Hispanics, and whites are 25.8%, 31.8%, and 27.6.
Insurance coverage
Inclusion of variables indicating the method of payment for the patient’s visit helps in determining the importance of insurance coverage on prescribing patterns of physicians.12 Type of insurance coverage is important in the case of coronary heart disease, where the coefficients of Medicare and Medicaid were negative and statistically significant at the 5 percent level. Patients whose main source of payment was Medicaid or Medicare were less likely to be prescribed a statin compared to patients with private insurance. None of the other payment variables appear to be important explanatory factors in the prescribing process for either statins or non-statin lipid-lowering drugs. For samples including individuals diagnosed with dyslipidemia and diabetes, the insurance 12 Also to note is the likelihood that these insurance data are reflective of more than a price variable. Sincesocioeconomic data are not available in the NAMCS, method of payment may also be picking up effects offactors such as family income and education of the patient.
variables were never significant determinants of whether or not a patient received a statin other than for unknown insurance where the patient was less likely than a privately insurance patient to receive a statin. As shown in table 3, the probability of a CHD patient being prescribed a statin in 2004 varies from 54.8 percent for those having private insurance to 47.6 percent for those in the Medicare category and 38.3 percent for those Secondary conditions
The presence of co-morbidities is stressed in clinical guidelines as an important factor affecting decisions to prescribe lipid lowering drugs. The presence of a secondary condition has a positive and statistically significant impact on the probability of being prescribed a statin in 19 out of 32 combinations. Although the remaining coefficients are not statistically significant, they are with three exceptions positive. 13 Table 3 shows the probabilities associated with the presence or absence of secondary conditions for the samples with coronary heart disease, dyslipidemia, and diabetes respectively. The probabilities of being prescribed a statin in 2004 are calculated with all variables held at their means except for time (= 2004) and the secondary condition indicated. The probability standing out for CHD patients occurs when dyslipidemia is also present. In this case the probability rises to 75.4% (compared to the mean for 2004 of 50.3%). Patients with dyslipidemia who have a secondary condition of CHD, hypertension, or diabetes each have a probability of about 62% of being prescribed a statin (compared to the probability at the means of 51.6%). Diabetics at the mean values of the variables in 2004 have a probability of 28% of being prescribed 13 The only negative coefficients never approached statistical significance.
a statin. When the person had a secondary condition of CHD, dyslipidemia, or dyslipidemia and CHD combined, the probabilities jumped to 44.9%, 57.9%, and 80.1% respectively. All of these results support the clinical guidelines that persons with primary conditions of CHD, dyslipidemia, and diabetes should be treated more aggressively than those not having secondary conditions. Patients diagnosed with a primary condition of CHD, dyslipidemia, or diabetes but no secondary condition, have a significantly lower probability of receiving a statin drug, 44%, 45.8%, and 19.4% respectively.
Geographic variation
The logit models reveal geographic disparities in prescribing patterns in the case of CHD and dyslipidemia. The probability of being prescribed a statin tends to be lower in all regions compared to the Northeast. CHD Patients residing in the Midwest and South are also less likely to be prescribed other lipid lowering drugs than similarly situated patients in the Northeast. Obviously, no clinical guidelines suggest that a patient living in one region of the country should be more likely to receive statins. These disparities add further support to the already documented differences in prescribing patterns across the United States (Burton, 2001); such disparities no longer come as a surprise, although they continue to foster concerns.
4. Summary and Conclusions
Coronary heart disease is the leading cause of death in the United States and high levels of LDL cholesterol, along with diabetes, are risk factors for cardiovascular disease.
This, along with the fact that statin therapy lowers LDL cholesterol and is beneficial to patients at risk for heart disease, including diabetics, led us to examine the prescribing patterns of physicians for samples of patients who have primary diagnoses of coronary heart disease, dyslipidemia, and diabetes. Obviously, the benefits from statins can be realized only if physicians are prescribing them to patients at risk of CHD.
Our results provide further evidence of racial/ethnic disparities, in this instance with respect for prescription drugs, since Blacks and Hispanics diagnosed with the primary condition of coronary heart disease (CHD) are less likely to be prescribed a statin. The predicted probabilities in 2004 for Blacks, Hispanics and whites are respectively 45%, 43.3%, and 51% respectively. This result is troubling but consistent with the results of the recent study conducted by the Institute of Medicine (IOM), which found that in many aspects of health care, including cardiac, minorities receive lower quality care compared to whites. The IOM study found persistent racial/ethnic differences in care even after controlling for such factors as access to the system, insurance status, and socioeconomic status.
Another major result is that CHD patients covered by Medicare or Medicaid were less likely to have been prescribed a statin compared to their counterparts who have private insurance. Medicaid (Medicare) patients were 16.5 (7.2) percentage points less likely to receive a statin compared to similarly situated persons having private health Patients having a risk factor along with their primary diagnosis were more likely to receive a statin than those without a secondary condition. For example, the probability is 75.4% for coronary heart patients who also have dyslipidemia. Diabetics exhibiting co-morbidities of coronary heart disease or a combination of both coronary heart disease and dyslipidemia have probabilities of 44.9 and 80.1 percent, respectively.
While the probability of being prescribed a statin has increased significantly over time for persons of all races and insurance types, the disparity in probabilities across races and insurance types has widened considerably. If this trend continues, non-white and publicly insured persons will trail by an ever greater margin white and privately insured persons for access to statins. Because of the racial issues associated with health insurance as well as access to care, equity in regards to prescription drugs must be a major issue in discussions of health care reform in the U.S.
REFERENCES
<http://www.diabetes.org/type-1-diabetes.jsp>.
<http://www.diabetes.org/type-1-diabetes/treatment-conditions.jsp>.
<http://www.diabetes.org/type-1-diabetes/complications.jsp>.
<http://www.diabetes.org/type-2-diabetes.jsp>.
<http://www.diabetes.org/diabetes-symptoms.jsp>.
<http://www.diabetes.org/diabetes-heart-disease-stroke.jsp>.
<http://www.diabetes.org/diabetes-cholesterol.jsp>.
American Heart Association (2005), Heart Disease and Stroke Statistics—2005 Update, Dallas, Texas: American Heart Association. (2007), “Heart Disease and Stroke Statistics -- 2007 Update At-a-Glance,” American Heart.org/statistics, Dallas, Texas: American Heart Association. 1 On 1 Health: Diabetes Basics.
<http://www.1on1health.com/1on1HealthWeb/controller/ArticleDisplayView/To picID=1009/LanguageID=1000/ AssetID=010002?state=110000000>.
Burton T. M. (2001), “Study Finds Prescription Drug Use Varies Widely from State to State.” The Wall Street Journal. June 19.
Centers for Disease Control and Prevention (CDC), U.S Department of Health and Human Services (2006a), “Chronic Disease – Preventing Heart Disease and (http://apps.nccd.cdc.gov), Atlanta, GA.
Centers for Disease Control and Prevention (CDC), U.S Department of Health and Human Services, (2006b), Division for Heart Disease and Stroke Prevention, “Addressing the Nations’ Leading Killers, 2006, Strategies for a Heart-Healthy and Stroke-Free America,” Centers for Disease and Control and Prevention (http://apps.nccd.cdc.gov), Atlanta, GA.
Centers for Disease Control and Prevention (CDC) (2003), “National Diabetes Fact Sheet: General Information and National Estimates on Diabetes in the United States, 2003,” Atlanta, GA: U.S Department of Health and Human Services.
Centers for Disease Control and Prevention (CDC) (2005), “National Diabetes Fact Sheet: General Information and National Estimates on Diabetes in the United States, 2005”, Atlanta, GA: U.S Department of Health and Human Services.
Cunningham, P.J., and May, Jessica H. (2006) “Medicaid Patients Increasingly Concentrated Among Physicians.” Tracking Report 16, Center for Studying Health System Change. Washington D.C.
DeNoon, Daniel (2003), WebMD Health: “Statins Cut Diabetes Heart Risk.” <http://my.webmd.com/content/article/66/79805.htm>.
The Doctor’s Lounge (2003), “Diabetics Should Receive Statins, New Study Says,” <http://www.thedoctorslounge.net/pharmalounge/articles/statins_diabetes/>.
Doty, Michelle M., Holmgren, Alyssa L. (2006), “Health Care Disconnect: Gaps in Coverage and Care for Minority Adults: Findings from the Commonwealth Fund Biennial Health Insurance Survey (2005).” The Commonwealth Fund, August Inglehart, John K. ed. (2005), Health Affairs 24 (2).
Hebert, P.R., Gaziano, J.M., et al (1997) “Cholesterol Lowering with Statin Drugs, Risk of Stroke, and Total Mortality: An Overview of Randomized Trials.” Journal of the American Medical Association 278(4): 313-321.
Kaiser Family Foundation, American College of Cardiology Foundation (2002), Racial/Ethnic Differences in Cardiac Care: The Weight of the Evidence. Menlo Park, CA: Henry J. Kaiser Family Foundation.
Knopp, R.H. (1999), “Drug Treatment of Lipid Disorders”. New England Journal of LaRosa, J.C., He, J., Vupputuri, S. (1999), “Effect of Statins on Risk of Coronary Disease. A Meta-analysis of Randomized Control Trials.” Journal of the American Medical Association 282(24): 2340-2346.
Llanos, Karen and Palmer, Lindsay (2006), “Using Data on Race and Ethnicity to Improve Health Care Quality for Medicaid Beneficiaries,” Center for Health Care Strategies, Inc., Lawrenceville, NJ.
Manning, Anita (2004), “Statins Gain Wonder Status: But Dangers Mean Drug Is Not For Everyone,” USA Today, June 29, 2004, p. D1.
Mayo Clinic (2005), http://www.MayoClinic.com/health/highcholesterol/AN01079) Mayo Clinic website (2007), http://www.mayoclinic.com/health/coronary-artery- Pearson, Thomas A., (2007) “The Prevention of Cardiovascular Disease: Have We Really Made Progress?” Health Affairs, 26 (1), pp. 49-60. Majumdar, Sumit R, Gurwitz, Jerry.H., Steven B. Soumerai, (1999) “Undertreatment of Hyperlipidemia in the Secondary Prevention of Coronary Artery Disease.” Journal of General Internal Medicine 14: 711-717.
Noonan, David (2003), “You Want Statins with That?” Newsweek, July 14, 2003, pp. 48- Smedley, B. D., Stith A.Y., Nelson, A.R. eds. (2002), Unequal Treatment: Confronting Racial and Ethnic Disparities in Health Care, Institute of Medicine of the National Academies. Washington, DC: National Academies Press.
Schappert S.M., & Nelson C.R. (1999), National Ambulatory Medical Care Survey, 1995-96 Summary, National Center for Health Statistics, Vital Health Stat 13 Szabo, Liz. “Hospital Inequalities Widen the Care Gap,” USA Today, October 25, 2005, U.S Department of Health and Human Services (2000), “1999 NAMCS Micro-Data File Documentation,” Rockville, MD: National Center for Health Statistics.
Table 1: Selected Sample Means by Condition, NAMCS, 1992-2004 Table 2: Multinomial Logit Estimates for the Probability that a Statin is Prescribed by a Physician, by Medical Condition of the Patient, 1992-2004 (Standard errors in ** = Significant at the 5% level.
* = Significant at the 10% level. No coronary heart disease sufferers who are prescribedan other lipid altering drug reported self pay insurance, resulting in a fixed parameter forthe combination of chd/olad/self pay.
Table 3: Predicted Probabilities Associated with a Patient Being Prescribed a Statin in Characteristics
*The first (second) row is calculated at the means for all variables in 1992 (2004). Allother rows are for 2004, and assume that all variables but the one indicated in the row ofinterest are at their mean values for 2004.

Source: http://www.lerner.udel.edu/sites/default/files/ECON/PDFs/RePEc/dlw/WorkingPapers/2008/UDWP2008-07.pdf