|Home | About | Journals | Submit | Contact Us | Français|
A usual source of care (USOC) has been associated with improved preventive and chronic care, but its relationship with lipid management has not been well described. The objective of this study is: 1) to examine the association of USOC with statin use among persons meeting eligibility guidelines for treatment 2) to examine the association of USOC with low density lipoprotein goal attainment among those receiving statins.
We examined statin use among adults ages 21-79 years who participated in the National Health and Nutrition Examination Survey data, 1999-2006. We used criteria from the third Adult Treatment Panel (ATP III) on treatment of high cholesterol to assess eligibility and the examined factors that predicted current use of statins. Among those currently taking statins, we assessed achievement of target LDL-C based on ATP III goals.
Among the 12,979 participants, nearly 14% were eligible. Having a USOC was significantly associated with use of statins among those eligible, but not with goal attainment among those taking statins. Significant predictors of goal attainment were higher income and education and more recent cohort.
In a national sample, USOC was significantly associated with use of statins among eligible adults, but not with LDL-C goal attainment for those using statins.
Having a usual source of care (e.g. first contact) is a core feature of primary care and is embedded in the principles of the patient-centered medical home.1 Having a usual source of care has been associated with improved access to care including improved receipt of preventive care,2,3weight loss among patient screened for hypertension or cholesterol, communication,4 adherence, and health status.5 When usual source of care is combined provider accessibility and well organized care, key health care disparities are eliminated.6 Disruptions in usual source of care are associated with emergency department visits7 and avoidable hospitalizations.8 Usual source of care is also associated with better chronic care management and disease control for HIV, asthma, hypertension, and diabetes.9-13
Yet, there are few data regarding the impact of continuity of care on cholesterol management beyond improvements in cholesterol screening.14,15 Using recent national data, we examined two questions: 1) Is continuity of care associated with appropriate treatment of high cholesterol with statins and 2) Is continuity of care associated with attainment of low density lipoprotein goals, among persons currently treated with statins?
We used publicly available data from the National Health and Nutrition Examination Survey (NHANES) conducted from 1999–2000, 2001-2002, 2003-2004, 2005-2006 by the National Center for Health Statistics of the Centers for Disease Control and Prevention. NHANES is an ongoing survey designed to provide nationally representative estimates for the non-institutionalized population of the United States based on a multistage, stratified sampling frame.16 Survey data include household interviews, examinations, and testing. Following the interview, participants are invited to mobile examination centers. The protocol for each NHANES was approved by the National Center for Health Statistics of the Centers for Disease Control and Prevention Institutional Review Board. Informed consent was obtained from each participant.
Our sample was restricted to the 16,681 adults (over 20 years) for whom ATP III criteria are developed (that is, excluding those ≥ 80 years). LDL-C values were available for 16,021 persons and, of these, complete data for all study variables was available on 12,979 (81.0%) persons corresponding to 84.4% of the target population (adjusting for population weights).
This was assessed by response to two questions:1) Is there a place that you usually go when you are sick or you need advice about health? 2) What kind of place do you go most often? We coded participants as having a usual source of care based on an affirmative response to the former question and reporting going to “a clinic or health center”, “doctor’s office or HMO”, or “hospital outpatient department.”
self reported age, sex, race, ethnicity, socio-economic status [percent of federal poverty], and educational attainment),
(LDL, HDL, total cholesterol, blood pressure, history of coronary heart disease (CHD), myocardial infarction, stroke, angina, diabetes, cigarette smoking, use of antihypertensive medication, family history of CHD).
Language usually spoken at home (English or not), presence of health insurance (any or none), number of visits in past year (intervals)
Participants were classified as having hypertension if, based on the average of 3 blood pressure measurements, they had a systolic blood pressure of 140 mm Hg or more and/or diastolic blood pressure of 90 mm Hg or more and/or they reported currently using antihypertensive medication.
(including use of statins) was based on a series of questions about any prescription drugs the participant reported having taken during the previous month. For each drug reported, the participant was asked to show the medication container for the interviewer to record the drug name. This information was used to ascertain participants’ current use of any statin.
We estimated pretreatment TC, LDL-C, and HDL-C based on the expected reduction associated with the specific lipid lower therapy the patient was using (based on medication bottle inspection).17,18 In sensitivity analyses, we examined values ± 35% of the original estimated percent reduction. (The estimated percentage reduction for different lipid values associated with each lipid lowering drug are available upon request.) Because LDL-C values were only available on a subsample, we imputed LDL-C values based on TC and HDL-C using a predictive model developed using the subsample. The R square for LDL-C prediction was 0.90 and effect sizes for usual source of care (and other factors) were similar between the full sample and the subsample for whom LDL-C values were obtained.
We applied the Third Adult Treatment Panel (ATP III) recommendations updated in 2004 for treatment of high cholesterol in adults to assess participant eligibility for statin use.19,20 ATP III guidelines use variable low density lipoprotein cholesterol thresholds for treatment based on three categories of 10-year risk for CHD (<10%, 10-20%, >20%). Patients with no CHD or CHD risk equivalents (diabetes, stroke or peripheral vascular disease) are assessed for major CHD risk factors CHD risk (cigarette smoking, hypertension, low HDL-C (<40 mg/dl), family history of premature CHD, and older age (45≥ years for men; 55≥ y for women). HDL-C 60 mg/dl or greater is considered protective and results in loss of the equivalent of a major CHD risk factor.
Patients with ≤ 1 major CHD risk factor are considered to be in the lowest risk category with a 10 year <10%. Participants with two or more risk factors undergo Framingham risk scoring (FRS) in order to sort them into the same three risk group categories. Participants with CHD, CHD risk equivalents, or FRS >20% are placed in the highest risk group.
For adults with 0-1 risk factors, the statin eligibility threshold for elevated LDL-C is ≥ 190 mg/dl with a LDL-C goal of <160 mg/dl. For adults in the intermediate risk category (2+ risk factors) the goal is 130 mg/dl, with a statin threshold of 130 mg/dl for those with a 10-year risk of 10-20% and 160 mg/dl for those with a 10-year risk of <10%. For adults in the highest risk group, the statin threshold is ≥130 mg/dl with a goal of ≤ 100 mg/dl. In the 2004 update, an option was added for reducing the LDL-C < 70 mg/dl for those in the highest risk group.
The data were analyzed using SAS-Callable SUDAAN v10.0.0 (Research Triangle Park, NC) and incorporated the complex survey design specifications to yield appropriate standard errors and parameter estimates reflecting the non-institutionalized civilian US population. The dependent variable for the first logistic regression model was statin use among those eligible for statin therapy according to ATP III criteria and estimated pretreatment lipids. Social risk factors included age group (< 35, 35-44, 45-54, 55-64, and ≥65), gender, race/ethnicity (white, black, Hispanic, and other), years of schooling (<12 years, 12 years, > 12 years), percent of Federal poverty level (<100%, 100-199%, 200-299%, 300-499%, ≥500%). Access factors included having a usual source of care, insurance (any vs. none), and preferred language (English or other).
The dependent variable for the second set of logistic regression models was whether the LDL-C level was at goal, and included only those respondents currently taking statins; the independent variables were the same as those in the first set of analyses. In both analyses, two-way interactions among independent variables were explored. In this study, we focus on social risk factors. We used 0.05 (or no overlap with 1.0 for 95% Confidence Intervals [CI]) for tests of statistical significance.
Of the 12,979 persons with complete data, 2,524 (19.4%) reported no regular source of care. The characteristics of persons with a usual source of care compared to those without are shown in Table 1. Those without a usual source of care were younger, more likely male, more likely to be a race other than White, more likely to prefer a language other than English, poorer, less educated, more likely to be uninsured, have lower CHD risk and less likely to be taking statins. Each of these difference was statistically significant (p <0.0001).
Among the12,979 participants, 1,927 were potentially eligible for statin treatment. Among these, 25% were taking statins. After controlling for patient sociodemographic characteristics and access, those eligible with a usual source of care were significantly more likely (odd ratio [OR] 4.47; 95% Confidence Interval [CI]; 2.09- 9.54) to be on statins (Table 2). Other predictors of statin use included age and later survey year (i.e. statin use improved over time). These findings were robust to sensitivity analyses. Use of a lower threshold for high risk patients, e.g. <70 mg/dl of LDL-C. The results were similar in subsamples for whom LDL-C were availables or for whom ankle brachial blood pressure ratio was available and peripheral arterial disease (< 0.9 ) was included as a CHD risk equivalent. When we varied the estimated effect of cholesterol-lowering drugs on lipids by ±35% (of the original effect), the effect of usual source of care ranged from 2.0 to 2.9.
Among the 1,333 participants who were appropriately treated with statins, there was a trend towards association of usual source of care with achievement of LDL-C goal (OR 2.0 ;95% CI 0.94-4.6) that did not reach statistical significance. Significant predictors of failing to reach LDL-C goal including poverty, lower education and participation in early years. included being female, poverty and lower income (Table 3). Restriction of sample to participants on statins presumed eligible or to those for whom LDL-C values were available yielded similar findings.
In this nationally representative sample, we found that having a usual source of care was associated with appropriate treatment with statins. We also found a non-statistically significant trend for an association between usual source of care and attainment of LDL-C goals among patients appropriately treated with statins. Interestingly, socioeconomic factors (i.e. income and education) were not associated use of statins, but were associated with LDL-C goal attainment. Our findings are consistent with prior studies showing that having a usual source of care is associated with improved chronic disease management.9-13
Our findings for goal attainment are consistent with previous data showing that higher socioeconomic status is associated with improved quality of care,21-23 in part due to fewer cost barriers for higher SES persons.24 These difference may reflect differences in provider intensification of treatment or patient differences in adherence to treatment and/or therapeutic life style.25,26 Higher SES is associated with a healthy diet, physical activity and higher rates of obesity.27,28 Each of these factors might contribute to lower goal attainment by SES. Improvements in goal attainment over time has also been previously noted.29
Given the powerful effects of statins on cardiovascular and all-cause mortality,30,31 these results suggest that those lacking a usual source of care may be at higher risk. Specifically, less optimal management of high cholesterol among those without a usual source of care may contribute to higher long-term mortality.
Prior studies suggest that more than 20% of the adult population lack a usual source of care,32 slightly higher than the 19% in this sample. Consistent with previous studies,32,33 we found that being uninsured, Hispanic, less educated and poor were associated with no usual source of care. In our sample, nearly 50% of those without a usual source of care were uninsured and 45% had household incomes 200% of poverty.
The most frequent reasons cited in past studies for lacking a usual source of care include lack of perceived need (i.e. never/seldom being sick) and cost associated with care.32 The reasons for lack of perceived need are unclear, but may reflect lower illness burden and lack of perceived benefit from preventive care. Costs are a well-recognized barrier to insurance.
Having a usual source of care is often a prerequisite for initiating and continuing chronic disease management. This may be particularly true for cholesterol management. An emergency department or urgent care physician is unlikely to assess a person’s need for statins or to start statin treatment. Similarly, patients who experience an interruption in usual source of care may no longer be able to obtain a prescription for their statin treatment.
Our findings are subject to several limitations. First, our measure of usual of source of care did not distinguish between primary or specialty care or provider continuity. Provider continuity benefits beyond usual source of care alone.2,3,34 It is conceivable that stronger effects might be observed with better measures. Second, our data are cross-sectional. We had no data on pretreatment cholesterol levels, drug doses, or number of times that physicians intensified therapy. We estimated pretreatment lipid levels based on expected reductions in lipids with different lipid lowering drugs. Nonetheless, our findings regarding the association of usual source of care with appropriate treatment with statins were robust in sensitivity analysis in which we varied the percent reduction in cholesterol with treatment. Given the relatively small subsample of participants for whom LDL-C was obtained, we imputed LDL-C based on TC and HDL-C. Comparison of effect sizes between the full sample and subsample showed similar results. The absence of an association between usual source of care and LDL-C was unexpected. This may reflect limitations of the measures discussed above and sample size limitations. Further research is needed to clarify confirm this. Previous studies have noted that African Americans are less likely to receive statins and reach goal attainment.29,35,36 Whether these differences reflect diffusion of statin across groups over time or potential bias in imputation of pretreatment LDL-C levels cannot be determined from these data.
In conclusion, these national data show that lower SES persons and men who are not on statins are more likely to be eligible and among those on statins, these groups are less likely attain their LDL-C goal. Further study is needed to determine the reasons and possible remedies for these disparities.
Funding: The National Heart Lung and Blood Institute (1R01 HL081066-01A2).