Statin Prescribing Patterns: An Analysis of Data From Patients With Diabetes in the National Hospital Ambulatory Medical Care Survey Outpatient Department and National Ambulatory Medical Care Survey Databases, 2005–2010

Clinical Therapeutics/Volume 37, Number 6, 2015

Statin Prescribing Patterns: An Analysis of Data From Patients With Diabetes in the National Hospital Ambulatory Medical Care Survey Outpatient Department and National Ambulatory Medical Care Survey Databases, 2005–2010 Brandy R. Pauff, BS; Michael R. Jiroutek, DrPH, MS; Melissa A. Holland, PharmD, MSCR; and Beth S. Sutton, PhD College of Pharmacy and Health Sciences, Campbell University, Buies Creek, North Carolina ABSTRACT Purpose: In 2008, the American Diabetes Association (ADA) recommended that patients aged 440 years with diabetes and cardiovascular disease or with Z1 cardiovascular disease risk factor be prescribed a statin. This study assessed statin prescribing patterns in patients with diabetes, per the ADA guideline, using data from the National Ambulatory Medical Care Survey and National Hospital Ambulatory Medical Care Survey–Outpatient Department for the years 2005 to 2010. This study also examined patients’ demographic characteristics associated with statin prescribing, including sex, age, ethnicity, race, insurance type, body mass index, region, primary care provider, hypertension and hyperlipidemia. Methods: This retrospective, cross-sectional, observational study included data dated between 2005 and 2010 from patients aged Z18 years with diabetes and without contraindications to statin use. Associations between statin prescribing and variables of interest were analyzed using χ2 tests. A multivariate logistic regression model included 2 groups stratified by 3-year observation period (2005–2007 and 2008– 2010) plus all variables with an overall χ2 test result of P o 0.2. P values, odds ratios (ORs) and 95% CIs are reported. Findings: The majority of patients were aged Z40 years (93.1%), had a body mass index of Z30 (58.7%), had hypertension (65.6%), and did not have hyperlipidemia (54.0%). A low percentage of patients were prescribed a statin (35.1%), but it appears that this percentage is on the rise. During 2005–2007, 31.9% of patients received a statin, whereas 37.7% of patients received a statin during 2008–2010. After adjustment for covariates included in the multivariate logistic regression model, those with hypertension

June 2015

(vs none [reference]: OR ¼ 1.31; 95% CI, 1.12– 1.53) and/or hyperlipidemia (vs none [reference]: OR ¼ 4.44; 95% CI, 3.70–5.33) were significantly more likely to have been prescribed a statin, whereas those in age group 18–o40 years (vs 40–o65 years [reference]: OR ¼ 0.45; 95% CI, 0.31–0.65) and Hispanic/Latino patients (vs non-Hispanic/Latino patients [reference]: OR ¼ 0.77; 95% CI, 0.61–0.97) were significantly less likely to have been prescribed a statin. Implications: Despite the call in the latest ADA recommendations for prescribing statins in many diabetic patients, an unexpectedly low percentage of patients were receiving them. Health disparities in age and ethnicity were also evident. The findings from this study highlight the need for further research into low statin prescribing rates. (Clin Ther. 2015;37:1329–1339) & 2015 Elsevier HS Journals, Inc. All rights reserved. Key words: diabetes, disparities, ethnicity, statins.

INTRODUCTION Diabetes mellitus continues to be a prominent disease across the world. An estimated 285 million adults worldwide had diabetes in 2010, and that number is expected to rise to 439 million by 2030.1 Diabetes results from a deficiency in the production and/or The data from this article were previously presented in abstract format at the American Diabetes Association’s 74th Scientific Sessions, June 13–17, 2014, San Francisco, California, and at regional universityaffiliated research symposiums. Accepted for publication March 11, 2015. http://dx.doi.org/10.1016/j.clinthera.2015.03.020 0149-2918/$ - see front matter & 2015 Elsevier HS Journals, Inc. All rights reserved.

1329

Clinical Therapeutics action of insulin.2 This deficiency leads to high levels of glucose in the blood, which can cause serious health consequences, such as kidney failure, blindness, heart disease, lower-limb amputations, and death.2–5 Patients with diabetes are at risk for cardiovascular diseases (CVD) resulting from hyperlipidemia, elevated levels of lipids in the blood. Atherosclerosis, the accumulation of lipids on the vessel walls, obstructs blood flow and oxygen delivery to the tissues, contributing to CVD.6 To control lipid levels and decrease CVD risk in patients with diabetes, hydroxymethylglutaryl–coenzyme A reductase inhibitors, or “statins,” are prescribed.7,8 Three important studies of the efficacy of statins in reducing hyperlipidemia in patients with diabetes have been conducted: CARE (Cholesterol and Recurrent Events),9 HPS (Heart Protection Study), 10 and 4S (Scandinavian Simvastatin Survival Study).11 The results from these studies have had an impact on guidelines on diabetes care. Before 2004, the American Diabetes Association (ADA) suggested that prescribers treat patients with diabetes and hyperlipidemia on an individualized basis.12,13 However, in 2004, the ADA guideline began to reflect the findings from the landmark studies, recommending that those with diabetes who are 440 years of age with a total cholesterol concentration of Z135 mg/dL be prescribed a statin.14 The ADA continued to update the guideline in 2005, recommending that all patients with diabetes and overt CVD and patients aged 440 with a total cholesterol concentration of Z135 mg/dL be prescribed a statin.12–20 In 2008, the ADA guideline was clarified to recommend that all patients with diabetes and CVD and patients aged 440 years with diabetes and Z1 CVD risk factor be prescribed a statin.18 Major risk factors for CVD mentioned by the ADA include cigarette smoking, hypertension (systolic/diastolic blood pressure Z140/Z90 mm Hg), low high-density lipoprotein cholesterol (o40 mg/dL), a family history of premature ischemic heart disease, and age 440 years.18 One study has suggested differences in the prescribing of statin therapies based on a variety of patients’ demographic characteristics. In a retrospective (2002– 2004) study of data from patients with diabetes identified using the National Hospital Ambulatory Medical Care Survey (NHAMCS) and the National Ambulatory Medical Care Survey (NAMCS),

1330

including inpatient and outpatient departments, Segars et al21 reported several significant associations between sex, age, presence of hyperlipidemia, and statin prescribing. They reported that men were 38% more likely to have been prescribed a statin than were women, and that diabetic patients with hyperlipidemia were 45-fold more likely to have been prescribed a statin than were those who did not have hyperlipidemia.21 Patients with diabetes between the ages of 1 and 24 years were 0.1-fold as likely to have been prescribed a statin as those in the 45- to 64–year age group; patients 25 to 44 years of age were 0.48-fold as likely to have been prescribed a statin as those 45 to 64 years of age; and those who were 65 to 74 years of age were 1.38-fold as likely to have been prescribed a statin as those 45 to 64 years of age.21 The study did not report any significant differences in the prescribing of statins based on patients’ race, ethnicity, region, or insurance type. Overall, o25% of visits included a statin prescription.21 Due to the findings from Segars et al21 regarding differences in statin prescribing rates during the years 2002 to 2004, and the availability of more recent NHAMCS and NAMCS data, the goals of this study were to determine whether there have been any changes in statin-prescribing rates since 2004 and to identify differences in statin-prescribing patterns in terms of diabetic patients’ demographic characteristics (sex, race, age, ethnicity), socioeconomic characteristics (payment type, region, primary care provider (PCP)), and risk factors (tobacco use, body mass index [BMI], hypertension, hyperlipidemia).

PATIENTS AND METHODS This retrospective, cross-sectional, observational study analyzed data dated from 2005 to 2010 collected from the NHAMCS-Outpatient Department (OPD) or NAMCS database. The NHAMCS is a national probability sample of ambulatory visits made to nonfederal, general, and short-stay hospitals in the United States, conducted by the Centers for Disease Control and Prevention, National Center for Health Statistics (NCHS).22,23 Although the survey includes visits to selected emergency care departments, the present analysis focused solely on the visits to hospital OPDs. The survey has been conducted annually since 1992. The multistaged sample design was composed of 3 stages in the OPD component: 12 geographic primary

Volume 37 Number 6

B.R. Pauff et al. sampling units (PSUs) that comprised a probability subsample of PSUs from the 1985 to 1994 National Health Interview Surveys;  550 hospitals within PSUs; and patient visits within outpatient service areas. In these surveys, sample hospitals are randomly assigned to 1 of 16 panels that rotate across sixteen 4-week reporting periods throughout the year. The initial sample frame of hospitals was based on the 1991 SMG hospital database (now maintained by IMS Health Inc, Danbury, Connecticut).22–24 Hospitals are inducted into the NHAMCS by field representatives of the US Census Bureau. 22–24 Hospital staff or Census Bureau field representatives complete a patient-record form for each sampled visit based on information obtained from the medical record.22–24 The data collected include patients’ demographic characteristics, reasons for visit, vital sign measurements, cause(s) of injury, diagnoses rendered, diagnostic tests ordered, procedures provided, medications prescribed, providers consulted, and disposition including hospital discharge information if admitted (since 2005). 22–24 For the purposes of the present study, on average, for the years 2005 to 2010, 84% of hospitals sampled were in scope and had eligible OPDs. Of the sample clinics drawn from these hospitals,  87% responded fully or adequately, yielding an estimated overall average unweighted 2-stage sampling response rate of 73%. The NAMCS is an annual, national probability sample of visits made to the offices of non–federally employed physicians classified by the American Medical Association or the American Osteopathic Association as providing “office-based, patient care.”22–24 Physicians in the specialties of anesthesiology, pathology, and radiology are excluded.22–24 Further details on the types of contact excluded can be found at http://www.cdc.gov/nchs/ahcd/ahcd_scope.htm#namcs_ scope. The survey was conducted annually from 1973 to 1981, in 1985, and from 1989 to the present.22–24 The multistaged sample design is composed of a 3 stages: 112 geographic PSUs, composed of counties, groups of counties, and county equivalents or towns and townships within the 50 states and the District of Columbia; a probability sample of practicing physicians, with eligible physicians stratified by specialty; and office visits within the annual practices of sample physicians.22–24 The total physician sample is divided into 52 random subsamples (approximately) equal in

June 2015

size, with each subsample randomly assigned to 1 of the 52 weeks in a year.22–24 Each physician then systematically selects a random sample of visits from the reporting week.24 The US Census Bureau is the field data-collection agent for the NAMCS, but data collection is expected to be carried out by each physician or his/her staff.24 However, in practice, data collection is more often than not performed by the Census Bureau’s field representatives.24 Physicians are instructed to keep a daily listing of all patient visits (both scheduled and unscheduled, but not cancellations or no-shows) during the assigned reporting week.24 A random sample of these logged visits is then selected for inclusion in the database.24 The data collected include patients’ symptoms, diagnoses, medications, procedures, planned treatment, and demographic characteristics.24 On average, for the years 2005 to 2010, 67% of physicians sampled met the criteria required for database eligibility. Of the eligible (inscope) physicians, the average unweighted response rate was  60%. Both the NHAMCS and NAMCS are approved annually by the Ethics Review Board of the NCHS, with waivers of the requirements to obtain informed consent from patients and patients’ authorization of the release of medical-record data by health care providers.22–24 For both databases, data processing, including all medical and drug coding, are performed by SRA International, Inc (Durham, North Carolina) and are subjected to quality-control procedures. Rates of keying and coding errors on various survey items typically range between 0% and 1%.22–24 NHAMCS-OPD and NAMCS datasets covering 6 years (2005–2010) were included in the present study. Data from patients from either database who were 18 years of age or older and having a diagnosis of diabetes (International Classification of Diseases, Ninth Revision [ICD-9] codes 249.00–250.93) were included in the final analysis dataset, whereas patients with any of the following contraindications for statin use were excluded based on ICD-9 codes for any of the 3 available diagnostic fields: cirrhosis (571.2, 571.5, 571.6), liver disease (571.8, 571.9, 572.8, 751.62), hepatitis (070.0, 070.9, 091.62, 130.5, 571.1–573.3), alcoholism (303.00–303.03, V11.3), and pregnancy (V22.2). Across all 6 years included in the present study, 10,887 raw records from the NHAMCS-OPD database and 7645 from

1331

Clinical Therapeutics the NAMCS database met the inclusion criteria (18,532 combined). The survey data were analyzed using the sampled visit weight, that is, the product of the corresponding sampling fractions at each stage in the sample design. The sampling weights were adjusted by NCHS for survey nonresponse as appropriate within each database, yielding a nonbiased national estimate of visit occurrences, percentages, and characteristics. Both the NHAMCS and NAMCS utilize multistage probability designs with samples from PSUs. The cluster, stratum, and weighting variables (CPSUM [the clustered PSUs marker], CSTRATM [the clustered PSUs stratum marker], and PATWT [the patient visit weight], respectively) are the same in both the NHAMCS and NAMCS datasets. When combining data from the NHAMCS-OPD and NAMCS datasets, these same variables were utilized. Because of the complex sample design, sampling errors were determined using the SAS SURVEYFREQ and SURVEYLOGISTIC procedures (SAS Institute Inc, Cary, North Carolina), which took into account the clustered nature of the sample.25 The appropriate NOMCAR and DOMAIN statements/options were implemented in these procedures, as recommended by the NCHS. The dependent variable of interest was statin prescribing, where the denominator was the number of cases meeting the inclusion criteria. Statin prescriptions were identified using the appropriate medication codes found in any of the MED1 to MED8 or DRUGID1 to DRUGID8 medication fields. A Rao-Scott χ2 test of homogeneity was used for determining whether the proportions of adult patients with diabetes who received a statin prescription differed between 2 groups stratified by 3-year observation period (2005–2007 vs 2008–2010). A series of Rao-Scott χ2 tests of association were used for determining whether, in adult patients with diabetes, any association exists between statin prescribing and any of the following variables: sex, age, race, ethnicity, insurance type, region, BMI, tobacco use, PCP, and hypertension and hyperlipidemia statuses. PCP variable was “Yes” if the patient saw their PCP at the recorded visit, or “No” if the patient saw a physician other than their PCP at the recorded visit. These variables were grouped for analysis as follows: sex (male or female), age (18–o40, 40–o65, and Z65 years), race (white, black, other), ethnicity (Hispanic/

1332

Latino or non-Hispanic/Latino), insurance type (private insurance, Medicare, Medicaid, other), region (South, Midwest, West, Northeast), BMI (o18.5, 18.5–o25, 25–o30, and Z30 kg/m2), tobacco use (yes or no), and PCP (yes or no). Tobacco use was excluded from all analyses due to high levels of missing data. Odds ratio (ORs), corresponding 95% CIs, and P values are reported. Additionally, for all of the previously mentioned variables with 42 levels and an overall χ2 test result of P o 0.05, pairwise comparisons of interest were conducted to determine the variables in which significant differences were found. A multivariate logistic regression model was also constructed to evaluate the joint association between all of the independent variables of interest and the receipt of a statin prescription. As a primary model filter, only variables with an overall χ2 test of association P o 0.2 (year group was included regardless) were included in the multivariate model. ORs with corresponding 95% CIs and P values for each level of each variable included in the model (in comparison to each variable’s reference group) were reported. All analyses were generated using SAS version 9.3, utilizing the SURVEYFREQ and SURVEYLOGISTIC procedures.25 Per NCHS recommendations, any variable with a survey estimate based on either o30 records, a relative SE of 430%, or 430% missing data was excluded from the analyses due to potential unreliability. As this was a retrospective, hypothesis generating–type of study, no adjustments for multiple comparisons were made, and P o 0.05 was considered statistically significant.

RESULTS The raw sample size meeting the inclusion criteria totaled 18,532 patients. Weighting and clustering was accounted for to reflect national estimates in all of the following results. The mean (SE) BMI was 32.8 (0.20) kg/m2. The mean (SE) age was 62.0 (0.24) years. Most of the patients were from the NAMCS database (89.7%), visited clinics during 2008–2010 (54.7%), were female (52.7%), were between 40 and o65 years of age (47.8%), were non-Hispanic/Latino (85.4%), and were white (76.9%). More patients had Medicare (42.2%) than any other type of insurance, reflecting in part the 45.3% of the patients Z65 years of age who made up a significant portion of the

Volume 37 Number 6

B.R. Pauff et al. study population. The majority of patients (68.4%) saw their PCPs, and 39.4% of patients were located in the southern United States. Most of the patients (58.7%) were obese (BMI Z30 kg/m2), with an additional 26.0% being overweight (BMI 25–o30 kg/m2). The majority of patients (65.6%) had hypertension, did not have hyperlipidemia (54.0%), and had not received a prescription for a statin (64.9%). This information is summarized in Table I. In the primary (weighted) analysis, during 2008– 2010, patients were significantly more likely to have been prescribed a statin compared with patients during 2005–2007 (OR ¼ 1.29; 95% CI, 1.10–1.52; P ¼ 0.0017). In 2008–2010, 37.7% of patients received a statin prescription, whereas 31.9% of patients in 2005–2007 received a statin prescription (Table II). In the secondary (weighted) χ2 analyses of other covariates of interest, the following variables were significantly associated with statin prescribing: age, sex, race, ethnicity, insurance type, PCP, hypertension, and hyperlipidemia. Patients aged 18 to o40 years were 0.38-fold as likely to have been prescribed a statin as those aged 40 to o65 years (16.8% vs 34.7%; 95% CI, 0.27–0.53; P o 0.0001). Women were 0.84-fold as likely to have been prescribed a statin as men (33.2% vs 37.1%; 95% CI, 0.75–0.94; P ¼ 0.0029). Black patients were 0.84-fold as likely to have been prescribed a statin as white patients (31.5% vs 35.4%; 95% CI, 0.70–1.00; P ¼ 0.0444). Hispanic/Latino patients were 0.74-fold as likely to have been prescribed a statin as those who are non-Hispanic/Latino (29.0% vs 35.4%; 95% CI, 0.62–0.90; P ¼ 0.0019). Patients with insurance from a Medicaid/State Children’s Health Insurance Program were 0.72-fold as likely as those with private insurance to have been prescribed a statin (28.4% vs 35.4%; 95% CI, 0.58–0.91; P ¼ 0.0044). Patients who did not see their PCPs were 0.71-fold as likely to have been prescribed a statin as those who saw their PCPs (30.1% vs 37.8%; 95% CI, 0.58–0.87; P ¼ 0.0008). Patients who had hypertension were 1.98-fold as likely to have been prescribed a statin as those who did not have hypertension (40.2% vs 25.3%; 95% CI, 1.75–2.24; P o 0.0001). Patients who had hyperlipidemia were 4.56-fold as likely to have been prescribed a statin as those who did not have hyperlipidemia (53.0% vs 19.8%; 95% CI, 3.86– 5.39; P o 0.0001) (Table II).

June 2015

Table I. Baseline characteristics of the patients in this analysis of statin prescribing patterns (N ¼ 18,532). Characteristic Database NHAMCS-OPD NAMCS Observation period 2008–2010 2005–2007 Sex Female Male Age Mean (SE) Group 18–o40 y 40–o65 y Z65 y Ethnicity Hispanic/Latino Non-Hispanic/Latino Race White Black Other Insurance Medicare Private insurance Medicaid/SCHIP Other BMI† Mean (SE) Group o18.5 kg/m2 18.5–o25 kg/m2 25–o30 kg/m2 Z30 kg/m2 Region Northeast Midwest West South No PCP Hypertension Hyperlipidemia Statin prescription

No. (%) of Patients*

33,358,983 (10.3) 289,488,029 (89.7) 176,513,238 (54.7) 146,333,774 (45.3) 170,201,331 (52.7) 152,645,681 (47.3) 62.0 (0.24) 22,049,254 (6.8) 154,425,146 (47.8) 146,372,612 (45.3) 38,082,107 (14.6) 221,938,731 (85.4) 201,763,698 (76.9) 43,860,881 (16.7) 16,682,068 (6.4) 131,459,313 121,777,491 37,916,153 20,444,768

(42.2) (39.1) (12.2) (6.6)

32.8 (0.20) 2,219,904 22,500,478 42,096,702 95,102,609

(1.4) (13.9) (26.0) (58.7)

62,337,204 76,017,817 57,282,195 127,209,796 96,867,998 211,800,821 148,507,138 113,181,280

(19.3) (23.5) (17.7) (39.4) (31.6) (65.6) (46.0) (35.1)

BMI ¼ body mass index; NHAMCS-OPD ¼ National Hospital Ambulatory Medical Care Survey–Outpatient Department; NAMCS ¼ National Ambulatory Medical Care Survey; PCP ¼ primary care provider; SCHIP ¼ State Children’s Health Insurance Program. * Values are appropriately weighted and clustered to reflect national estimates. † BMI o18.5 group: only 15 patients in this BMI group were prescribed a statin.

1333

Univariate Parameter

Volume 37 Number 6

Observation period 2008–2010 2005–2007 Age group 18–o40 y Z65 y 40–o65 y Sex Female Male Race Black Other White Ethnicity Hispanic/Latino Non-Hispanic/Latino Insurance Medicare Medicaid/SCHIP Other Private insurance BMI‡ o18.5 kg/m2 25–o30 kg/m2 Z30 kg/m2 18.5–o25 kg/m2 Region Northeast Midwest

Multivariate †

OR (95% CI)

P

0.0017 —

1.18 (0.97–1.43) Reference

0.1035 —

0.38 (0.27–0.53) 1.16 (1.03–1.31) Reference

o0.0001 0.0178 —

0.45 (0.31–0.65) 1.06 (0.88–1.26) Reference

o0.0001 0.5587 —

113,675,021 (66.8) 95,990,711 (62.9)

0.84 (0.75–0.94) Reference

0.0029 —

0.90 (0.80–1.02) Reference

0.0957 —

13,805,236 (31.5) 6,029,498 (36.1) 71,512,485 (35.4)

30,055,645 (68.5) 10,652,570 (63.9) 130,251,213 (64.6)

0.84 (0.70–1.00) 1.03 (0.74–1.44) Reference

0.0444 0.8569 —

0.96 (0.76–1.22) 0.99 (0.69–1.42) Reference

0.7475 0.9457 —

11,043,548 (29.0) 78,601,259 (35.4)

27,038,559 (71.0) 143,337,472 (64.6)

0.74 (0.62–0.90) Reference

0.0019 —

0.77 (0.61–0.97) Reference

0.0234 —

Statin, No. (%)

No Statin No. (%)

OR (95% CI)

66,543,410 (37.7) 46,637,870 (31.9)

109,969,828 (62.3) 99,695,904 (68.1)

1.29 (1.10–1.52) Reference

3,706,410 (16.8) 55,855,498 (38.2) 53,619,372 (34.7)

18,342,844 (83.2) 90,517,114 (61.8) 100,805,774 (65.3)

56,526,310 (33.2) 56,654,970 (37.1)

P

48,851,963 10,761,987 7,086,000 43,087,267

(37.2) (28.4) (34.7) (35.4)

82,607,350 27,154,166 13,358,768 78,690,224

(62.8) (71.6) (65.3) (64.6)

1.08 (0.94–1.25) 0.72 (0.58–0.91) 0.97 (0.75–1.25) Reference

0.2917 0.0044 0.8077 —

1.00 (0.83–1.20) 0.93 (0.72–1.20) 1.16 (0.83–1.61) Reference

0.9601 0.5548 0.3848 —

900,412 16,672,718 37,570,066 8,153,193

(40.6) (39.6) (39.5) (36.2)

1,319,492 25,423,984 57,532,543 14,347,285

(59.4) (60.4) (60.5) (63.8)

1.20 (0.58–2.51) 1.15 (0.86–1.54) 1.15 (0.86–1.53) Reference

0.6187 0.3312 0.3408 —

— — — —

— — — —

39,409,364 (63.2) 48,473,780 (63.8)

1.14 (0.91–1.43) 1.12 (0.88–1.41)

0.2450 0.3565

— —

— —

22,927,840 (36.8) 27,544,037 (36.2)

(continued)

Clinical Therapeutics

1334

Table II. Statin prescriptions, univariate and multivariate analysis.* Data are given as number (%) of patients.

June 2015

Table II. (continued). Univariate Parameter West South Risk factors PCP No Yes Hypertension Yes No Hyperlipidemia Yes No

Multivariate †

OR (95% CI)

P

0.7286 —

— —

— —

0.71 (0.58–0.87) Reference

0.0008 —

0.89 (0.70–1.13) Reference

0.3249 —

126,714,391 (59.8) 82,951,341 (74.7)

1.98 (1.75–2.24) Reference

o0.0001 —

1.31 (1.12–1.53) Reference

0.0007 —

69,853,079 (47.0) 139,812,653 (80.2)

4.56 (3.86–5.39) Reference

o0.0001 —

4.44 (3.70–5.33) Reference

o0.0001 —

Statin, No. (%)

No Statin No. (%)

OR (95% CI)

19,798,137 (34.6) 42,911,266 (33.7)

37,484,058 (65.4) 84,298,530 (66.3)

1.04 (0.84–1.28) Reference

29,197,576 (30.1) 79,163,720 (37.8)

67,670,422 (69.9) 130,429,972 (62.2)

85,086,430 (40.2) 28,094,850 (25.3) 78,654,059 (53.0) 34,527,221 (19.8)

P

BMI ¼ body mass index; OR ¼ odds ratio; PCP ¼ primary care provider; SCHIP ¼ State Children’s Health Insurance Program. BMI and region were not included in the multivariate model. See the methods section for details. * Values are appropriately weighted and clustered to reflect national estimates. † 2 χ Analysis. ‡ BMI o18.5 group: only 15 patients in this BMI group were prescribed a statin.

B.R. Pauff et al.

1335

Clinical Therapeutics In the (weighted) multivariate logistic regression analysis, only the following variables retained significance as predictors of statin prescribing: age (18–o40 years vs 40–o65 years), ethnicity, hypertension, and hyperlipidemia. After adjustment for the covariates included in the model, patients were significantly less likely to have been prescribed a statin if they were aged 18 to o40 years versus 40 to o65 years (OR ¼ 0.45; 95% CI, 0.31–0.65; P o 0.0001) or were Hispanic/Latino versus non-Hispanic/Latino (OR ¼ 0.77; 95% CI, 0.61–0.97; P ¼ 0.0234). Patients were significantly more likely to have been prescribed a statin if they had hypertension (OR ¼ 1.31; 95% CI, 1.12–1.53; P ¼ 0.0007) or hyperlipidemia (OR ¼ 4.44; 95% CI, 3.70–5.33; P o 0.0001) compared with those who did not (Table II).

DISCUSSION

Segars et al21 reported that during o25% of visits in 2002 to 2004, a statin was prescribed, but the present study found that 31.9% of the 2005–2007 cohort received a statin prescription compared with 37.7% in the 2008–2010 cohort. Although the overall statin prescribing rate appears to be increasing, the prescribing rate in adults with diabetes remains low overall, conflicting with current ADA guidelines and recommendations. A significant difference between year groups in the proportions of patients who received a statin prescription was detected using an individual χ2 test in the present study, but this difference was not significant in the multivariate model, suggesting only a modest change in statin prescribing rates across time when other factors were taken into consideration. Prescribers’ perceptions regarding whether statin use leads to increases in hemoglobin A1c and fasting plasma glucose, possibly causing diabetes, may contribute to low statin prescribing rates. The findings from a cost–benefit analysis of statin use in patients with diabetes but without CVD is still debated; in that analysis of data from the National Health and Nutrition Examination Survey, it was less costeffective to treat patients aged 40 to 75 years with diabetes and no CVD, than to treat patients in the same age group with diabetes and a 47.5% risk for CVD over a 10-year period.26 Research published soon after the 2005–2010 study period reported conflicting results as to whether statin use increases hemoglobin A1c and fasting plasma glucose

1336

concnetrations,27 but more recent research has reported that antihyperlipidemic agents seem to protect patients with CVD risk against mortality.28 Even with ADA recommendations, some physicians may be hesitant to prescribe a statin if they believe that these adverse events (AEs) are possible. Additionally, the following AEs have been shown to be associated with statin use: muscle pain, tenderness, and weakness (myopathy and myalgia); rhabdomyolysis; increased serum aminotransferases such as creatinine phosphokinase; and decreased steroid hormones.29–34 During 2004–2009, 2.49% of the myalgias, rhabdomyolysis, and elevated creatinine phosphokinase AEs reported to the US Food and Drug Administration were associated with statin use.35 The inconsistent results published in the literature, along with the generally agreed-on statinassociated AEs, may also be contributing to prescribing rates lower than those expected with the existing guidelines. Perhaps not surprisingly, the 18 to o40-year age group (vs 40–o65 years) was significantly less likely to have been prescribed a statin after adjustments for other covariates of interest. Similar findings on age were noted in the study by Segars et al,21 as diabetes was, and remains, more prevalent in older patients. What is more, this finding is consistent with ADA guideline recommendations of a statin prescription for every patient aged 440 years with diabetes and at least 1 CVD risk factor.12–20 However, in the present study, in the 40 to o65-year age group, 34.7% were prescribed a statin, and in the Z65-year age group, 38.2% were prescribed a statin. Although the number of CVD risk factors was not considered in the present study, these percentages were expected to have been higher in the present study population of adults with diabetes, as 93.1% of the patients were aged Z40 years, 58.7% were obese, 65.6% had hypertension, and 46.0% had hyperlipidemia. These percentages suggest that patients with diabetes and CVD risk factors are not receiving statin prescriptions in accordance with the ADA guidelines. The study by Segars et al21 reported that patients with hyperlipidemia were significantly more likely to have been prescribed a statin; this finding remained significant in our multivariate model. In addition, hypertension remained a significant factor in our multivariate model of statin prescribing. Segars et al21 did not evaluate hypertension, despite

Volume 37 Number 6

B.R. Pauff et al. hypertension being a known risk factor for CVD.36 However, despite these significant associations, once again lower-than-expected percentages (53.0% and 40.2%, respectively) of patients with the CVD risk factors of hyperlipidemia and hypertension were prescribed a statin. The authors’ concerns over disparities in statin prescribing were for the most part alleviated. Although age group, sex, race, ethnicity, insurance type, and PCP were variables showing significant associations in the individual χ2 tests, only the associations between statin prescribing and age group and ethnicity remained significant in the multivariate model. As discussed previously, the significantly decreased likelihood of the 18 to o40-year age group receiving a statin prescription was perhaps not surprising due to the higher rate of diabetes in the older age groups and the ADA guideline. However, Hispanic/Latino patients appear to remain significantly less likely to receive a statin prescription. This finding was not expected because 1 study found that Mexican Americans had significantly higher hemoglobin A1c levels, took significantly more oral antidiabetic agents, and had a significant intensification of drug regimen in the previous year, indicating a less well-controlled disease state, compared with non-Hispanic white patients.37 Although this research was conducted using (although not exact) variables and methods similar to those in the study by Segars et al,21 data on physician specialty and degree type were not available from all years and both databases; thus, these variables were unable to be analyzed. However, Segars et al21 did not identify an association with these variables. In addition to the variables that Segars et al21 analyzed, several key risk factors for CVD, including hypertension, hyperlipidemia, and obesity, were examined. Additional CVD risk factors were not analyzed due to limitations of data availability from the databases. Tobacco use was originally planned to have been analyzed; however, due to small numbers, the data were deemed unreliable and thus are not reported. Moreover, Segars et al21 did not include an age limitation, whereas the present study included only adult diabetic patients, thus limiting potential confounding information due to differences in pediatric and adult patients. Furthermore, Segars et al21 used NHAMCS–Emergency Department

June 2015

data in the analyses, whereas the present study focused only on the OPD section of NHAMCS, thus focusing only on ambulatory care in the United States. The cross-sectional nature of the present study allowed for the analysis of data only from individuals at the moment in time at which they participated in the survey, eliminating any chance to collect repeated measurements on patients over time. Although this study included data on hypertension, hyperlipidemia, BMI, and tobacco use, additional CVD risk factors such as low high-density lipoprotein cholesterol were not captured in the databases and therefore were unable to be analyzed.15 Additional variables that were available only through ICD-9 codes (eg, family history and CVD) were not analyzed due to the limitation of only 3 diagnostic fields in the databases. The NHAMCSOPD and NAMCS surveys have a limited number of diagnoses and medications recorded, so it is possible that the diagnosis of diabetes or statin use may not have been included. Also, it was not possible to determine from the available data whether statin myopathy or other AEs were considered as reasons for the lack of a statin prescription. Compared with the study by Segars et al,21 which was based on 3 years of data, from 2002 to 2004, the present study included 6 more recently available years of data, 2005 to 2010. By combining 6 years of data from 2 large-scale, national, validated surveys, an increased sample size was obtained, providing additional generalizability and increased accuracy of the results. Furthermore, the surveys in both databases were completed by the provider, rather than the patient, allowing for less bias and increased accuracy.

CONCLUSIONS After adjustment for covariates, age, ethnicity, hypertension, and hyperlipidemia were significantly associated with statin prescribing in patients with diabetes. This study highlights the need for further research into low statin prescribing rates, particularly with regard to ethnicity and CVD risk factors.

ACKNOWLEDGMENTS We thank Victoria Lancaster for her assistance with the protocol design and data analysis. All of the authors conducted protocol design and data analysis, reviewed/edited the manuscript, and

1337

Clinical Therapeutics approved the final version of the manuscript. Ms. Pauff and Dr. Jiroutek wrote the manuscript. None of the authors receive any current support nor have received any previous support from industry or other organizations that might have been considered to influence this work.

CONFLICTS OF INTEREST The authors have indicated that they have no conflicts of interest with regard to the content of this article.

REFERENCES 1. Wild S, Roglic G, Green A, et al. Global prevalence of diabetes: estimates for the year 2000 and projections for 2030. Diabetes Care. 2004;27:1047. 2. American Diabetes Association. Diagnosis and classification of diabetes mellitus. Diabetes Care. 2011;34:S62–S69. 3. Parving HH, Smidt UM, Andersen AR, Svendsen PA. Early aggressive antihypertensive treatment reduces rate of decline in kidney function in diabetic nephropathy. The Lancet. 1983;321:1175–1179. 4. Zhang X, Saaddine JB, Chou CF, et al. Prevalence of diabetic retinopathy in the United States, 20052008. JAMA. 2010;304:649–656. 5. Gu K, Cowie CC, Harris MI. Diabetes and decline in heart disease mortality in U.S. adults. JAMA. 1999;281:1291– 1297. 6. Kreisberg RA, Reusc JE. Hyperlipidemia (high blood fat). JCEM. 2005;90:0. 7. Last AR, Ference JD, Falleroni J. Pharmacologic treatment of hyperlipidemia. Am Fam Physician. 2011;84:551–558. 8. Parris ES, Lawrence DB, Mohn LA, Long LB. Adherence to statin therapy and LDL cholesterol goal attainment by patients with diabetes and dyslipidemia. Diabetes Care. 2005;28:595–599. 9. Plehn JF, Davis BR, Sacks FM, et al. Reduction of stroke incidence after myocardial infarction with pravastatin: the Cholesterol and Recurrent Events (CARE) study. Circulation. 1999;99:216–223. 10. Collins R, Armitage J, Parish S, et al, Heart Protection Study Collaborative Group. MRC/BHF Heart Protection Study of cholesterol-lowering with simvastatin in 5963 people with diabetes: a randomized placebo controlled trial. Lancet. 2003;361:2005. 11. Pyorala K, Pedersen TR, Kjekshus J, et al. the Scandinavian Simvastatin Survival Study (4S) Group. Cholesterol lowering with simvastatin improves prognosis of diabetic patients with coronary heart disease: a subgroup analysis of the Scandinavian Simvastatin Survival Study. Diabetes Care. 1997;20:614–620.

1338

12. American Diabetes Association. Standards of medical care for patients with diabetes mellitus. Diabetes Care. 2002; 25:213–229. 13. American Diabetes Association. Standards of medical care for patients with diabetes mellitus. Diabetes Care. 2003; 26:S33–S50. 14. American Diabetes Association. Standards of medical care in diabetes. Diabetes Care. 2004;27:S15–S35. 15. American Diabetes Association. Standards of medical care in diabetes. Diabetes Care. 2005;28:S4–S36. 16. American Diabetes Association. Standards of medical care in diabetes - 2006. Diabetes Care. 2006;29:S4–S42. 17. American Diabetes Association. Standards of medical care in diabetes–2007. Diabetes Care. 2007;30:S4–S41. 18. American Diabetes Association. Standards of medical care in diabetes–2008. Diabetes Care. 2008;31:S12–S54. 19. American Diabetes Association. Standards of medical care in diabetes–2009. Diabetes Care. 2009;32:S13–S61. 20. American Diabetes Association. Standards of Medical Care in Diabetes–2010. Diabetes Care. 2010;33: S11–S61. 21. Segars LW, Lea AR. Assessing prescriptions for statins in ambulatory diabetic patients in the United States: a national, cross-sectional study. Clin Ther. 2008;30:2159–2166. 22. McCraig LF, Burt CW. Understanding and interpreting the National Hospital Ambulatory Medical Care survey: key questions and answers. Ann Emerg Med. 2012; 60:716–721. 23. Centers for Disease Control and Prevention. Sample text for describing NHAMCS in a research article. http://www. cdc.gov/nchs/data/ahcd/Sample_Text_for_Describing_N HAMCS_in_Research_Article.pdf. Updated August 6, 2014. Accessed August 18, 2014. 24. National Center for Health Statistics. Scope and sample design. http://www.cdc.gov/nchs/ahcd/ahcd_scope.htm#namcs_ scope. Updated January 15, 2010. Accessed May 2, 2014. 25. SAS 9.3. Cary, NC: SAS Institute Inc, 2012. 26. Busko M. Statins not cost-effective for many newly eligible patients. Medscape. 2014. 27. Uchechukwu KS, MacRae FL, Fazzio S. Are statins diabetogenic? Curr Opin Cardiol. 2011;26:342–347. 28. Cox AJ, Hsu FC, Freedman BI, et al. Contributors to mortality in high-risk diabetic patients in the Diabetes Heart Study. Diabetes Care. 2014;37:2798–2803. 29. Crestor [package insert]. Wilmington, DE: AstraZeneca; February 2009. 30. Pravachol [package insert]. Princeton, NJ: Bristol-Myers Squibb Company; March 2007. 31. Mevacor [package insert]. Whitehouse Station, NJ: Merck & Co Inc; September 2008. 32. Lescol and Lescol XL [package insert]. East Hanover, NJ: Novartis; October 2006.

Volume 37 Number 6

B.R. Pauff et al. 33. Lipitor [package insert]. New York, NY: Parke-Davis; February 2009. 34. Zocor [package insert]. Whitehouse Station, NJ: Merck & Co Inc; June 2011. 35. Sakaeda T, Kadoyama K, Okuno Y. Statin-associated muscular and renal adverse events: data mining of the public version of the FDA adverse event reporting system. PLoS One. 2011;6:1–5. 36. Banjari I, Bajraktarovic-Labovic S, Huzjak B. Dietetic approaches in prevention and treatment of cardiovascular diseases. Acta Med Median. 2014;53:65–72. 37. Kaplan SH, Billimek J, Sorkin DH, et al. Reducing racial/ethnic disparities in diabetes: the Coached Care (R2D2C2) Project. JGIM. 2013;28:1340–1349.

Address correspondence to: Beth S. Sutton, PhD, Campbell University College of Pharmacy and Health Sciences, PO Box 1090 Buies Creek, NC 27506. E-mail: [email protected]

June 2015

1339