Effect of Insurance Type and Patient Compliance on Control of Cardiovascular Risk Factors in Diabetic Patients
Is pay-for-performance really fair to health care providers? This chart-study points to some answers
By William Ervine Jr., DO, William Burke, DO, and Richard Snow, DO
In 1998, the American Osteopathic Association (AOA) created the Clinical Assessment Program (CAP). CAP is a quality improvement program designed to help physicians improve patient care by comparing outcomes and clinical indicators for a number of disease states and preventive care in their practice with those of other physicians around the country. The areas being studied include diabetes mellitus, low back pain, hyper- tension and metabolic syndrome, women’s health screening, childhood immunizations, coronary artery disease and adult immunizations.
Some of the examples of indicators being investigated include aspirin use in coronary artery disease patients, chlamydia screening in all women under the age of 26 years and appropriate immunizations provided to a two year-old child. All osteopathic family practice residencies have been required to participate in the program since 2003. Starting in 2005, all osteopathic internal medicine programs were required to participate as well. As of June 30, 2004, 54 osteopathic family practice residency programs and three osteopathic internal medicine residency programs have participated in this program.
Presently, the AOA is integrating practicing physicians into a similar program to allow them to benefit from this quality improvement tool. Another important aspect of the CAP is that it examines the type of outcomes that are being used in pay-for-performance programs.(1, 2)
Pay-for-Performance Reimbursement
Pay-for-performance is the new reimbursement system for doctors and hospitals, which is meant to reward health care providers for delivering quality care. The premise for these systems is to provide increased reimbursement to those providers whose patients meet certain measures.
These measures look at utilization and cost management, clinical quality/effectiveness, patient satisfaction, administrative and patient safety. Many different programs have been proposed, and vary on the number of these measures and which specific end-points are used. Some may use utilization measures such as the percentage of generic drugs prescribed, the number of unnecessary magnetic resonance imaging or computerized tomography tests, and the average number of emergency department visits per patient per year. Other factors include administrative issues such as the level of information technology in a practice, like electronic medical records, and patient satisfaction by determining the percentage of patients that would recommend a doctor to others in their community or timely access to care. They may also factor in patient safety by looking at the percentage of patients questioned about drug side effects.
Yet, the most common measures being looked at focus on clinical quality/effectiveness measures. These measure end-points such as the percentage of asthmatic patients on controller medications, cervical cancer screening, mammography, immunizations, blood glucose testing in diabetics and low-density lipoprotein (LDL) cholesterol management in heart disease patients. Even specific outcomes such as blood glucose and LDL cholesterol levels are being used more frequently.(3, 4, 5)
All of these measures are assumed to be a reflection of the care provided by health care providers. However, the outcomes of health care are not solely a reflection of the health care worker’s ability or services provided. Outcomes in health care are based on three factors: access, utilization and quality of health services.
Access is related to whether patients have insurance and their insurance type. It may determine what doctors they can see, when they can see them, the amount of their co-pay and deductible, and what diagnostic and treatment options are available to them.
Utilization refers to the patient qualities, whether they go to the doctor, at what stage in their disease process they seek medical attention, and if they are compliant with the medical advice they are given.
Quality of health services refers to all of the variables related to health care providers and the care they provide.
Indeed, outcomes can reflect the care provided by physicians, but is a reimbursement system that only factors in the quality of health services without factoring access to health care and patient attributes truly fair?(6-11)
Materials and Methods
Data was obtained from the American Osteopathic Association’s (AOA) Clinical Assessment Program (CAP). The data contained in the CAP was obtained by retrospective chart review of randomly-selected charts in AOA-accredited residency programs. A total of 57 residency programs across the country have submitted data to the CAP, of which 54 are family practice programs and three are internal medicine programs.
A resident selects a certain disease process that the AOA has designated as one to be monitored for quality assurance purposes. The resident then reviews 20 to 40 charts of patients in his or her practice with the selected disease process. In doing so, he or she collects a number of demographic and outcome data points. These data points are then compiled into one database, which can then be used to give an overall rate of achieving certain goals in all of the residency programs, as well as in individual residency programs.
Data for diabetes mellitus was obtained from the academic years 2002-2003, from a total of 48 residency programs. A total of 2,047 patient charts were reviewed, and abstracted information submitted to the database. The demographic factors of age, gender, insurance type, number of kept and missed office visits, as well as outcome data of hemoglobin (Hgb) A1C and low-density lipoprotein (LDL) cholesterol were focused on in this study. A cohort of 1,334 patients more than 18 years of age was obtained by excluding any data sets that were missing data for age, gender, insurance type, number of kept and missed office visits, LDL cholesterol or Hgb A1C levels.
This cohort was then used to look for any clinically significant differences (p-value of 0.05) in the control of Hgb A1C and LDL cholesterol of diabetics based on the demographic factors. Hemoglobin A1C control was considered as seven percent or less, and low-density lipoprotein cholesterol control as 100 mg/dL or less as defined by the American Diabetes Association (ADA) and Adult Treatment Panel (ATP) III guidelines, respectively.(12, 13) Insurance status was categorized as Medicare, Medicaid, commercial insurance or self-pay/other. Gender was defined as male or female. Age was divided into those more than age 55, or those 55 years old and younger. The data for number of kept and missed office visits was changed into a fraction by dividing the number of missed office visits by the total number of office visits (kept plus missed). From this, patient non-compliance was selected to be represented by a fraction of 0.2 or greater. This meant non-compliance on the part of the patient was defined as missing one out of every five office visits or more. SAS/STAT Software was used for all statistical analyses. Simple T-tests were used to detect significant differences based on age, gender and patient non-compliance, while a chi-square analysis was used to determine significance based on insurance type.
Results
There were 1,334 patient charts included in this study. Overall, 20.2 percent of patients were classified as noncompliant, with 54 (14.3 percent), 76 (24.2 percent), 83 (19.1 percent) and 56 (27.1 percent) patients having Medicare, Medicaid, commercial and self-pay/other were classified as non-compliant, respectively. The total number of patients having Medicare, Medicaid, commercial insurance and self pay/other were 378 (28.3 percent), 314 (23.5 percent), 435 (32.6 percent) and 207 (15.5 percent), respectively. The number of male patients was 538 (40.3 percent) and the number of patients older than age 55 was 690 (51.7 percent).
The percent of patients with and without each attribute (insurance type, gender, age, compliance) obtaining control of HgbA1c and LDL cholesterol along with a p-value for having that particular attribute can be found in Table 1.
The overall rate of diabetes control and LDL control in this cohort was 42 percent and 41 percent, respectively. As seen in the table, a significant difference was seen in control of diabetes and LDL cholesterol if the healthcare payer was either Medicare (p-values 0.0014 and 0.0208, respectively) or self-pay/other (p-values 0.0016 and 0.0088, respectively).
Those patients with Medicare as the payor of their health care showed 48.68 percent and 46.03 percent obtaining control of their diabetes and LDL cholesterol, respectively, as opposed to 39.12 percent for both diabetes and LDL control in those without Medicare as their healthcare payer.
Conversely, those who were self-pay/other had lower rates of control of diabetes and LDL cholesterol at 31.88 percent and 32.85 percent, respectively, as opposed to 43.66 percent diabetes control and 42.59 percent LDL cholesterol control in those patients without that payor type. There was no significant difference if patients’ payor-type was Medicaid (p-values 0.0809 and 0.6929) and commercial insurance (p-values 0.3409 and 0.5772) in terms of diabetes and LDL cholesterol control, respectively. No significance was seen based on gender (p-values 0.4943 and 0.1737) or age above 55 years old (p-values 0.0876 and 0.0653) in terms of diabetes and LDL cholesterol control, respectively.
There was a significant difference in diabetes control (p-value 0.0104) and LDL cholesterol control (p-value 0.0221) if the patient was non-compliant with their office visits. Patients who were non-compliant controlled their diabetes and LDL cholesterol 34.94 percent for each, as opposed to 43.57 percent and 42.63 percent, respectively, if they were compliant with their office visits. In clinics that reported on more than 10 patients, the range in percent of self-insured patients was between zero and 52 percent.
Discussion
Some of the limitations of this study are that there could have been selection bias when the residents picked out charts to extract the data. Incomplete charting and documentation in those charts reviewed may have caused skewing of the data. Another limitation that occurs with retrospective chart reviews is errors in abstracting the data. These limitations may have skewed the data to either make it worse or better. Yet, they would probably not affect the significance of the data obtained. It was also not determined if the differences seen were correlated in any way, such as the non-compliance being significant because there were significantly more self-pay patients in that grouping, or vice-versa.
The cost of health care in the United States is growing every year, and there appears to be no end in sight. One of the concerns is that the consumer is not getting the quality health care that they are paying for and thus pay-for-performance was created to address this concern. Thus, pay for performance links the pay of the health care provider to the “quality of care” he or she provides. Unfortunately, the quality of care the physician provides can be excellent, yet the outcomes may not be up to the desired levels. And yet, there are some physicians that do not deliver the standard of care, who will be financially forced to improve the quality of care they provide.
As demonstrated in this study, the payor (Medicare, Medicaid, commercial insurance, self-pay) and patient compliance can have a significant effect on the outcomes. So it is potentially unfair to base a health care provider’s reimbursement on a system that does not factor in these other variables. Health care providers may want to get as many Medicare patients as they can into their practice, because they are more likely to have good outcomes. Conversely, health care providers may attempt to limit the number of self-pay patients in order to improve their outcomes data. Even though there was no significant difference in those with Medicaid, it did come close to having a negative effect on diabetes control. This may cause health care providers to be more hesitant about taking on Medicaid patients if they are going to be reimbursed less because of aspects that are not under their control.
It is also possible that patients who are non-compliant will also find it more difficult to receive health care when they do go to the doctor. In one residency clinic, it is policy that if a patient misses three visits in a year the patient is sent a letter stating that if one more scheduled visit is missed before a year has passed, the patient will be dismissed from the practice. This will be one step many physicians will have to take to financially insulate their practice, and many may have to lower the number of missed visits before patients are dismissed from their practice.
This will be unfortunate, because not only may it affect those patients who are non-compliant in other aspects of their health care, but also the patients who may have forgotten about the appointment or experienced unavoidable circumstances that prevented them from making their appointment. Many patients could suffer, as they will lose the continuity of care they receive from having a physician who they know and trust caring for them. This continuity of care would presumably lead to better quality of care than can be obtained from having a new health care provider every year or two, assuming the provider is competently following conventional treatment guidelines. Also, there were non-compliant patients in all payor groups with the largest number in the self-pay/other group. Thus, non-compliance is an issue for all insurance types, and must be factored into the reimbursement structure, i.e., pay-for-performance and quality programs.
Another concern that has been voiced in the literature is that pay-for-performance, with its incentives and payments based on outcomes, may drive health care providers away from caring for socially-disadvantaged patients. It has been found that this population of patients tends to be less compliant, have poorer support systems to help them with their chronic conditions, and often seek health care later in the disease process.(14) This would force health care providers to make a difficult decision between providing care to inner-city patients and other socially-disadvantaged patients while receiving less reimbursement, or seeing fewer of these patients and thus receiving better compensation for their time.
Although most health care providers are altruistic and would not completely abandon giving health care to the socially-disadvantaged, very few would set up their practice in these areas for any length of time. One of the major objectives of leaders in the health care field is to attract quality healthcare providers to work in areas where there are socially-disadvantaged patients, so that the patients will have greater access to health care. Yet pay-for-performance could have a detrimental effect on reaching this objective because of economic disincentives, unless resources are allowed for non-compliance, which many national programs have advocated, such as the Bridges to Excellence not-for-profit coalition.
The argument that health outcomes in areas with high concentrations of socially-disadvantaged patients are sub-par because of the lack of quality health care close to home has been put forth. Kirby and Kaneda disproved this argument when they found the supply of health care providers in disadvantaged areas did not explain why these patients were less likely to have a consistent source of health care and to obtain routine preventive services.(15-16)
Another issue will arise when outcomes data on every health care provider is published for the benefit of patients to choose their physician. The bias discussed above may skew the data for those physicians who take care of non-compliant and socially-disadvantaged patients, so that these physicians may lose compliant patients or not pick up any new compliant patients, as there are models to label them as non-referred, disability, co-pays and deductibles if patients see these providers. Therefore, health care providers who provide excellent health care may be punished financially with loss of quality established and new patients because of their patient population.
Our health care system needs to find a way to engage patients to take responsibility for their own health care. In general, it appears that our society seems to promote the idea that if there is a problem it has to be someone else’s fault. If a patient does not have control of his or her diabetes, it must be the health care provider’s fault. It could not possibly be that the patient does not take his or her medicine as they should, that they do not keep their scheduled appointments, or that they are gaining weight because they are not sticking to the recommended dietary changes and exercise routine. Patient accountability must be a factor, and some programs are beginning to factor this element in monthly premiums.
As a society and as a profession, we need to get patients to take more responsibility for their health care. This will not be an easy task.
Making non-compliant patients pay more for health insurance may not be practical because it would create another barrier to obtaining quality health care. Yet, this is happening in many states and private, commercial insurance products. Unfortunately, this may take an overhaul of societal thinking and living to do this. Things like fast food-restaurants, the foods found in grocery stores, television, video games, and tobacco and alcohol use would have to be addressed so that patients could be more compliant with their health care providers’ instructions.
Conclusion
Ultimately, our sense of entitlement as a society needs to change, because no one owes us anything and we need to work to get what we want, including quality health care. Lawyers, politicians and health insurance executives need to realize that we must start with the patients. When we get patients to take responsibility for their health care, then outcomes will improve and medical providers will be better able to deliver quality health care.
Pay-for-performance is not a bad idea, but rather, is one that may be perceived as unfair to health care providers as some programs presently proposed more refinement before implementation. Trying to find a way to get every individual involved in his or her health care is a good idea. There are a number of health care providers that do not always provide high quality health care. So it is a good idea to provide an incentive for physicians to provide quality health care, but not by using a system that would punish health care providers delivering quality health care to patients that are non-compliant or socially disadvantaged.
Any system that is meant to help patients receive quality health care needs to factor in the effects that the patient and the healthcare payer have on obtaining quality health care, as measured by outcome data.
Drs. Ervine, Burke, and Snow are members of the medical staff of Doctors Hospital, Columbus, Ohio.
| Table 1: Association between selected attributes and diabetes and LDL control | ||||||
| Patient Attribute (Number of Patients with Attribute) | Rate of HgbA 1C < 7% with Attribute | Rate of HgbA 1C < 7% without Attribute | p-value | Rate of LDL < 100 mg/dl with Attribute | Rate of LDL < 100 mg/dl without Attribute | p-value |
| Medicare (378) | 48.68% |
39.12% |
0.0014 |
46.03% |
39.12% |
0.0208 |
| Medicaid (314) | 37.58% |
43.14% |
0.0809 |
42.04% |
40.78% |
0.6929 |
| Commercial (435) | 43.68% |
40.93% |
0.3409 |
40.00% |
41.60% |
0.5772 |
| Self Pay /Other (207) | 31.88% |
43.66% |
0.0016 |
32.85% |
42.59% |
0.0088 |
| Male (538) | 40.71% |
42.59% |
0.4943 |
43.31% |
39.57% |
0.1737 |
| Missed >= 20% of visits (269) | 34.94% |
43.57% |
0.0104 |
34.94% |
42.63% |
0.0221 |
| Age > 55 years (690) | 44.06% |
39.44% |
0.0876 |
43.48% |
38.51% |
0.0653 |
References: