CONCEPTUAL FRAMEWORK
2.3 Adapted Model
ABM has been widely used to explain and predict health service use in health
economic and policy research.19 One remarkable application of the ABM on health policy
Personal health practices Lifestyle behaviors Adherence Use of Health Services Hospitalizations Outpatient visits Office visits Perceived Health Status Evaluated Health Status Consumer Satisfaction Healthcare System
Hospital beds per capita Primary physicians per capita Education level Poverty level Predisposing Demographics Age Sex Regions Marital status Social Education Race/ethnicity Health Belief Care-seeking attitudes Obesity Smoking Enabling Factors Social Support Financial Income OOP costs Part D (Cost-sharing) Need Perceived Self-perceived need ADL/IADL Evaluated Diagnosed conditions CONTEXTUAL CHARACTERI STICS INDIVIDUAL CHARACTERISTICS HEALTH BEHAVIOR Costs OUTCOME
Figure 2.3 Adapted Andersen’s Behavioral Model of Health Services
co-insurance on demand for health services.36,37,41, Furthermore, ABM includes a wide variety of factors influencing healthcare use, such as, demographics, socioeconomics, health behaviors, which are critical in understanding the dynamic relationship between insurance plan and health outcomes. As mentioned previously the theory of price elasticity of demand can also be adapted to ABM to illustrate the relationship between drug price and demand for prescriptions and non-drug medical services. Therefore, ABM can be applied to this dissertation to explain the effect of Part D plans on the use and costs of health services.
Figure 2.3 demonstrates the adapted version of ABM for this dissertation. When applying the ABM to this study, the adapted model focused on three domains: contextual characteristics, individual characteristics, and health behavior. Health outcome domain was not included in the adapted model, because the outcome measures of this dissertation are the healthcare utilizations and medication adherence, which have been covered under the health behavior domain. The key independent variable – type of part D plans – was considered as one of the individual enabling factors that have direct effects on the health behaviors, e.g., medication adherence and use of health services. With lower levels of generosity of Part D plans (e.g., PDPs), the demand for prescription drugs will decrease as the effect price of prescription drugs increases. This decreased use of prescription drugs may lead to lower medication adherence and possibly result in increased use of non-drug medical services (e.g., hospitalizations).
2.3.1Contextual characteristics
Several studies identify environmental or regional factors related to healthcare delivery and access. For each county code, we included 5 such factors reflecting
predisposing and enabling characteristics. Contextual predisposing factors were reflected by the education level, income and employment rate in the area. Two variables were used to reflect health system capacity, including the number of primary care physicians and hospital beds available in an area. Primary care physicians included family physicians, non-specialist internal medicine physicians, general practitioners, and general practice obstetricians and gynecologists. These factors reflect how health services facilities and personnel distributed to provide the services within an area and may influence
individual’s access to care, e.g., the availability of physicians/hospitals.29
2.3.2Individual characteristics
For the individual predisposing factors, demographic factors include age and gender, which may be associated with health care use. Social factors were reflected by education level, Race/ethnicity, living conditions, and geographic location. Higher levels of education may be associated with greater knowledge about care and more positive attitude in seeking care.42 Race/ethnicity and geographic location have been shown to be linked to different treatment patterns for specific conditions. Individual’s social
connections can also influence the access to care, and were measured by the living conditions. Health beliefs can be reflected by the care-seeking attitudes that affect
individual’s perception of need and use of health services. The measures included how an individual seeks care when sick, such as, avoidance of going to see a doctor, keeping to himself, visiting a physician as soon as he can. Differences in the attitudes toward seeking professional health care have been documented in the published studies.
Individual enabling factors were measured by income and the effective price of health services determined by having health insurance and co-insurance. People with
higher levels of income tend to use more health care, with all other factors being equal. As mentioned earlier, people purchase more health services as their prices decline. With lower percentage of coinsurance, people demand more health care as the effective price is lowered. The type of Part D plans, as the key independent variable of this dissertation, was included under this domain because it influences the effective price of prescriptions.
Individual need factors were reflected by both perceived and evaluated needs. Perceived needs were reflected by self-perceived health status, self-reported difficulties in Activities of Daily Living (ADLs) and Instrumental Activities of Daily Living
(IADLs). An individual with higher self-perceived health and function status is associated with lower uses of health services. On the other hand, evaluated needs were reflected by the presence of chronic conditions diagnosed by health care professionals.
2.3.3Health behavior
Personal health practices were measured as use of tobacco and presence of
obesity, which are proxy measures of an individual’s lifestyle. For example, an individual with normal weight is more likely to have healthy diet and exercise, compared to those with obesity.
This domain also included the primary outcome of interest of this dissertation – the actual use of health care (including prescriptions). I assumed that the contextual and individual characteristics mentioned above influence the use of health services. Since healthcare expenditures are highly correlated with health resource use, and healthcare costs were included under this domain.