A Description of Treatments

In document Three essays in health economics (Page 162-169)

Insurance Type Differences in Medical Productivity

4.4 A Description of Treatments

Diseases can vary significantly in their treatment complexity, protocols and health consequences. For instance, diabetes treatment is very complex, and may involve a lifetime of hospital visits, insulin shots, and other drug regimens. In contrast, upper respiratory infections can be very simple to treat and may involve a trip to the doctor’s office and a week’s worth of antibiotics. The response to treatment may also vary across diseases. The known conditions have a wide spectrum of potential outcomes that differ by disease. Diseases may be preventable with vaccine, curable,

Table 4.3: Per Person Disease Costs Condition Private Medicaid Uninsured

Depression 1001 1547 947

Hypertension 602 1061 587

Arthritis 656 684 513

Diabetes 1107 1879 1349

treated indefinitely but never cured, or have no available treatment. The differences in these outcomes have changed over time with technological advancement. Related to the potential outcomes of the treatment are the health consequences of disease treatment. Conditions may vary significantly in how they affect health. Some diseases may increase mortality risks but are not physically debilitating, such as high cholesterol. Other conditions may be physically debilitating but do not change the risk of death, such as arthritis. Other conditions, such as diabetes, are potentially both debilitating and mortal.

The treatment outcomes are related to the nature of the disease, and are consequently also very heterogeneous. Disease treatments may be able to address some of the undesirable properties of the disease, but not others. For instance, diuretic drugs may help with the uncomfortable aspects of hypertension, but may not address the mortality risks associated with the disease. In some cases medical care services may worsen some aspects of health while treating other aspects. For instance, some arthritic drugs have been shown to increase mortality, although they are effective at relieving pain and morbidities. Cancer treatment often employs therapies that increase morbidities in an attempt at lowering mortality.

4.4.1 Cost Per Disease

Although the nature of disease treatment is extremely complex, a good starting point for comparing the nature of them begins by comparing the average total costs per

person across diseases. The following analysis considers three disease treatments - arthritis treatment, depression treatment, and hypertension treatment. These conditions represent three of the ten most prevalent chronic conditions in the United States, and two of the ten most costly conditions to treat in the United States. All three conditions have effective treatments available that involve multiple medical care services.

The costs associated with a specific ICD-9 code include all of the expendi-tures paid to a medical care provider made by and in behalf of the patient with the condition. The expenditures included in this calculation are limited to those expenditures associated with the medical events that specifically list the three digit ICD-9 code as the reason for the event. Therefore, an individual who has both diabetes and hypertension and visits the physician for a checkup is included in the total cost of hypertension only if the checkup specifically lists hypertension as the reason for the visit.9 Medical events that are never associated with ICD-9 codes are not included in the per person cost of specific conditions. The costs of specific diseases, as determined by the ICD-9 code, are calculated by summing the total annual expenditures included in the treatment of the disease and dividing the sum by the total number of individuals receiving treatment for that condition within the year. All costs are associated with the calendar year in which they were accrued.

Individuals with a disease include only those individuals who had some positive spending on the disease within the year. 10 The costs in each year are weighted to reflect a nationally representative sample, and are deflated by the annual Consumer Price Index-Urban presented in year 2000 dollars.

The per person treatment costs of the conditions considered are economically

9The costs associated with events that list multiple conditions as the reason for the event are included in the treatment costs of all the conditions listed.

10This may significantly affect the cost per person, as some individuals may have a disease and receive treatment, but spend zero dollars on treatment because of bad debt or the receipt of charity care.

important.11 Diabetes treatment, the most expensive condition considered, is more than $1,000 per person per year. Arthritis treatment, the least expensive condition considered, is less than $700 per person per year. These differences in costs suggest that either the treatment complexity differs across diseases, or that the prices of the inputs used to treat the diseases are different, or both.

Comparison of treatment costs for the same condition across insurance types reveals some striking evidence in the cost per person of treating diseases. Table 3 reveals that on average, Medicaid insurance pays the most for disease treatments per patient per year for every condition considered. The uninsured pay less for treatment costs than does either the privately insured or Medicaid patients for all of the conditions considered except diabetes. The size of these differences can be quite large. Medicaid pays at least 50% more than does private insurance for all treatments other than arthritis treatment. Private insurance pays up to 30% more than the uninsured, but does pay less for diabetes treatment.

The reasons for these differences are not revealed by examining aggregate cost differences. Treatment cost differences across diseases and between insurance types have many potential sources including differences in treatment algorithms, severity of the conditions treated, and potential treatment compliance. In the case of the uninsured, the differences in costs may potentially reflect the provision of free or discounted care by service providers to the uninsured, or bad debts incurred and not paid in full during the survey period. Explaining the reason for the observed differences requires identifying these effects.

11The relative size of these cost differences is economically important but small relative to other medical conditions that could have been considered. In contrast to the cost differences observed between the conditions considered here, H.I.V. treatment can be upwards of $4,000 per person per year and upper respiratory infections may be less than $50 per person per year.

4.4.2 Disease Treatment Types

The reasons for why differences in the costs of disease treatments are observed is addressed by examining the medical care services used in the treatment of disease.

Protocol differences across insurance types are defined as input service differences used in disease treatment. Table 4 examines aggregate protocol and cost differences in disease treatments. The disease treatment protocols are defined as bundles of input services specific to treat a disease and are listed in Table 4 under the column heading ”Treatment”. The treatments are defined by whether individuals have some positive spending on a particular bundle of medical care service events associated with the disease. The treatments represent an exhaustive list of potential treatments available to the patient.

The number and types of treatment are specific to the disease but represent four broadly defined categories of services: hospital care, office visits alone, drugs alone and combinations of non-hospital services. A hospital treatment is defined as any treatment that includes a hospital stay or visits to either the emergency room or an outpatient facility located within the hospital.12 Treatments that in-clude office visits may involve visits to either physician or non-physician offices.

The drugs considered depend on the disease. Arthritis drugs include Non-steroidal anti-inflammatory medications such as Ibuprofen, narcotics such as codeine, and other analgesics such as Cox-2 inhibitors. Depression drugs include Selective Sero-tonin Re-uptake Inhibitors (SSRIs) such as Fluoxetine HCl (Prozac), and other anti-depressants such as Wellbutrin. Depression also considers anxiolytics, and anti-convulsant medications such as Diazepam. Hypertension drugs include Beta-adrenergic blocking agents (Beta-blockers) such as Atenolol, and Angiotensin Con-verting Enzyme (ACE) Inhibitors such as Lisinopril. Anti-hypertensive medications

12Events must be associated with an ICD-9 code for the expenditures to be included in the share calculations. For this reason, the caveats associated with how expenditures were allocated to cost per person calculations also apply to Table 3 as well.

Table 4.4: Treatment Costs and Utilization

Treatment Per Person Costs Utilization Rate 716 Arthropathies

Hospital 2756 0.095

Office Only 325 0.230

Office + Other Rx 802 0.090

Office + ”Ibuprofen” 463 0.136

Office + Other Rx 1055 0.113

SSRI Only 423 0.232

All Non-hospital 318 0.110

SSRI + Other Rx 1432 0.181

401 Hypertension

Office + Card Rxs 618 0.185

Office + Beta Blkr 917 0.045

Office + ACE + Card 839 0.049

Office + Rxs 809 0.103

Other Treatments 604 0.121

are often supplied in combination with each other and/or diuretics in a class of drugs referred to as ’Combination therapies’. Combination therapies are considered to-gether with ’other’ medications used to treat hypertension. The other medications may include other anti-hypertensive medications such as Amlodipine (a calcium channel blocker), or medications used to treat the confounding effects that hyper-tension has on other conditions such as high cholesterol. These drugs include the HMG-COA reductase inhibitors such as Atorvastatin. Chapter 2 provides a com-plete list of the items considered in each of the service categories.

The disease treatment technology is defined by the input services used to treat disease. The utilization rates that define the fraction of people who receive the specified treatment for the associated disease are presented in Table 4. The utilization rates suggest that the treatments are not very hospital intensive. Less than 10% of all treatments for any of these conditions involve hospital care. In contrast, at least 50% of all treatments involve an office visit, and more than 65%

of treatments involve the use of drugs. The intensity of these broad service types depend on the disease considered.

Treatments vary significantly in their costs within a condition. Hospital treatment stands out as being the most costly type of treatment. Hospital care is consistently the most expensive type of treatment for each of the diseases considered, and the costs of hospital treatment are up to 20 times higher than the costs of other types of treatments. Service combination therapy is typically more expensive than treatments that are intensive in only one type of service. Whether drug-only treatments or office-only treatments are more costly depends on the disease.

Tale 5 presents disease treatment costs by insurance types. Table 5 reveals that the costs of identical treatments for identical diseases varies across insurance types. The size of these differences can be dramatic. For instance, the privately insured hospital treatments for depression are $2500 dollars more expensive than

are hospital treatments for depressed Medicaid patients. Private insurance is not always more expensive than Medicaid insurance, however. Office visit treatments for hypertension can be close to $500 more expensive for Medicaid patients than are privately insured office visits.

Note that the direction of these ”costs” is inconsistent across insurance types.

Table 3 suggests that Medicaid patients are more costly to treat for identical con-ditions, but Table 5 suggests that the difference in these costs is not necessarily derived from the costs of individual treatments. Moreover, the observed differ-ences in the use and cost of these services by insurance types may represent quality differences, differences in demographic composition across insurance type popula-tions, the initial health of the individuals receiving treatment, or the behavioral differences of the patients receiving treatment. Identification of quality differences on health outcomes requires controlling for these other possibilities and measuring health outcomes without confounding the demographic and initial health effects with treatment effects. The evidence provided in Tables 1, 3 and 5 suggest that these differences are potentially important in determining treatments and outcomes.

In document Three essays in health economics (Page 162-169)