Disease Treatment Prices
3.3 Treatment Prices - A Disease Comparison
Disease treatment prices are constructed using the procedure described in Section 2 in order to control for service intensity, service type dynamics, and patient char-acteristics. This procedure implicitly uses index theory to construct a commodity aggregate for the price of a disease treatment. The commodities are medical care service inputs that are used in the treatment of a disease. The price of each medical care input service is constructed by controlling for the types of activities that occur during an event, demographic characteristics of the individual, and the insurance type that paid for the event. Service intensity is identified by constructing the share of expenditures spent on a specific service that is used in the treatment of a disease.
As section 3 has demonstrated, diseases vary dramatically in per person costs and the technologies used to treat them. These differences persist within aggregate categories of related diseases as defined by the ICD system. In order to account for these types of heterogeneity in the construction of disease prices the diseases are defined as narrowly as possible. The most specific definitions available on the MEPS are 3-digit ICD-9 codes. The 3-digit conditions considered are chosen from
each of the chapter headings. The chosen diseases are often either the most costly or most prevalent condition within each of the ICD-9 chapter headings.
3.3.1 Naive Estimates
Table 6 presents a summary of the treatment price increases for several diseases.
The results in Table 6 assume that treatment outcomes have not changed at all over the period. Presented along the rows are the average annual price changes over the period for each of the regions. For example, stomach infections in the Northeast have undergone a 33.6% increase that is equivalent to a 4.2% per year price increase averaged over the period 1996 to 2003.
Information concerning the total costs of the disease considered is also listed among the results presented in Table 6. This information includes the cost per person of the disease in 2000 dollars, the percent of total spending represented by this disease as a fraction of total ICD-9 chapter spending, and the annual average total costs of the disease in billions of dollars deflated to the year 2000. For example, stomach infections cost $1.29 billion per year, which is approximately $38 per person, and represents slightly more than 15% of total spending on infectious diseases. The percent of total spending represented by the disease considered is presented in order to provide a measure for how representative the disease is in each chapter.
Table 6 demonstrates that the rate of price increases are also very hetero-geneous across diseases. For example, consider three costly conditions: pregnancy, diabetes, and hypertension. These three conditions face vastly different price condi-tions. The rates of increase vary across diseases and the regions face different price increases across diseases. For example, diabetes faces the highest price increases in the Midwest, but hypertension has the highest rate of increase in the West and pregnancy faces the fastest rate of increase in the Northeast.
The most costly diseases have neither the lowest nor the highest rates of
Table 3.6: Summary of Disease Treatment Prices
Stomach Skin Depr- Otitis
Hyper-Variable Infections Cancer Diabetes ession Media tension
Chapter Infectious Neoplasms Endocrine Mental Nervous
Circul-Disease System Health system atory
Northeast Increase 0.0425 0.0782 0.0396 0.0316 0.0904 0.0378 Midwest Increase 0.0583 0.0543 0.0567 0.0437 0.0659 0.0390 Southeast Increase 0.0473 0.0537 0.0503 0.0371 0.0508 0.0392
West Increase 0.0446 0.0429 0.0521 0.0477 0.0451 0.0411
Non-MSA Increase 0.0407 0.0352 0.0398 0.0339 0.0414 0.0257 Annual Costs
(billions) 1.29 1.45 17.63 10.49 2.42 18.63
Cost per Person 38 957 531 159 446
% of Chapter 0.1531 0.0397 0.7478 0.3725 0.0650 0.2396
Preg-Variable AMI Disease Asthma Disorder nancy Arthritis
Chapter Circu- Circu- Respir- Gastro- Pregnancy
Musculo-latory latory atory Intestinal skeletal
Northeast Increase 0.0596 0.0642 0.0459 0.0402 0.0552 0.0423 Midwest Increase 0.0473 0.0419 0.0460 0.0250 0.0454 0.0408 Southeast Increase 0.0582 0.0458 0.0468 0.0384 0.0461 0.0378
West Increase 0.0605 0.0460 0.0437 0.0309 0.0448 0.0429
Non-MSA Increase 0.0271 0.0279 0.0440 0.0279 0.0235 0.0247 Annual Costs
(billions) 11.98 11.59 6.09 2.61 34.14 7.67
Cost per Person 7,067 1,429 436 281 4,959 388
% of Chapter 0.1540 0.1491 0.1643 0.0848 0.8859 0.1967
change over the period. Rather, Otitis Media faces the highest rate of price crease and stomach disorder treatment experiences some of the lowest price in-creases. Treatment prices for Otitis Media increase at a rate faster than 6.5% per year for two regions - the Northeast and the Midwest. No other conditions face rates of increase of at least 6.5% per year in more than one region. Moreover, Otitis Media prices in the Northeast increase at a rate of 9% per year which is 1.2% per year faster than the next highest rate of increase - skin cancer in the Northeast.
In contrast, Stomach disorders do not experience price increases greater than 4%
per year. Stomach disorder treatment prices in the Midwest face price increases as low as 2.5% per year. However, 2.5% per year is not the slowest rate of increase observed. Pregnancy services in the non-MSA region experiences a rate of increase less than 2.4% per year.
Overall, regional price differences and the magnitude of price increases de-pend on the disease, but some patterns appear. The Northeast appears to face some of the highest price increases for several conditions. The Northeast region has the highest or second-highest rate of increase for eight of the fourteen conditions con-sidered, which is greater than any other region. On the other hand, the non-MSA region never has the highest rate of increase and has the lowest rate of increase for nine of the fourteen conditions considered.
3.3.2 Accounting for Health Outcomes
Although the treatment prices presented in Table 6 account for heterogeneity in the types of services used to treat disease, they fail to account for whether changes in services are yielding different health outcomes over time. Failing to account for these changes implicitly ignores quality improvements of these services over time. The observed technological changes may be advancing the health of society, and an index of disease treatments should account for these health improvements. Accounting
for health improvements requires health outcome measures of each of the diseases considered.
Table 7 provides aggregate health outcome trends for several of the diseases considered. Under some (admittedly strong) assumptions, the association of the out-come measure with the disease whose treatment we consider suggests that changes in these measures over time identify the productivity of medical treatment.3 Simply stated, these measures represent medical care output.
The outcome measures considered are disease specific. For six conditions that include stomach infections, skin cancer, diabetes, acute myocardial infarction (heart attacks), heart disease, and hypertension we consider the age-adjusted mor-tality. The description for how the age-adjustment was performed for mortality is described in Section 2. Similarly, the age-adjusted suicide and infant mortality rates are used to evaluate depression and pregnancy, respectively. For all of the condi-tions considered here, mortality statistics are taken from vital statistics records and published by the Centers for Disease Control and Prevention (2005).
Asthma and arthritis use slightly different health measures. For asthma we use the percent of adults who experience an asthma attack within the past twelve months among self-reported asthmatics. Reported results are calculated and tabulated by the CDC (2005) from the Behavioral Risk Factor Surveillance Survey (BRFSS) over the period 1996-2002. For arthritis, the health measure is constructed by the author from the MEPS. The reported measure is the probability that an individual’s mobility will increase within the year conditional on initial health and demographic characteristics such as age, sex, race, income and insurance status.
Initial health is measured as whether an individual had any activity limitation prior to the initiation of arthritis treatment within the year. Mobility is measured as whether the individual walks a mile, walks up stairs, bends down, or grabs a pen
3see Heidenreich and McClellan (2001) for a discussion of this issue especially with respect to heart attacks.
Table 3.7: Health Outcomes
Disease 1996 1997 1998 1999 2000 2001 2002 2003
Age-adjusted Mortality Per 100,000
Stomach Infections - - - 0.37 0.45 0.51 0.77
-Skin Cancer - - - 3.4 3.4 3.5 3.4
-Diabetes 23.8 23.7 24.2 25.0 25.0 25.3 25.4 25.3
Depression (Suicide) 11.5 11.2 11.1 10.5 10.4 10.7 10.9 10.8
A.M.I. - - 71.8 69.2 67.1 63.9 60.5
-Heart Disease 212.1 203.6 196.9 194.6 186.8 177.8 170.8 162.9
Hypertension - - - 14.74 15.41 16.03 16.62
-Pregnancy (Infant 7.3 7.2 7.2 7.0 6.9 6.8 7.0
Asthma Attack within Last Year
Asthma 3.7 3.4 3.3 3.4 3.8 3.7 3.3
-Probability of Mobility Deterioration for Arthritics Arthritis 0.335 0.379 0.372 0.351 0.397 0.371 0.369 0.393
with difficulty. Mobility changes are measured as changes in the mobility status from the earliest wave in the year to the latest wave in the year.
Of the conditions that use mortality as the health outcome measure, stomach infections are the least deadly of all the conditions. Less than one person per 100,000 dies as a result of contracting a stomach infection. However, since 1999 the rate of mortality rate of this condition has more than doubled from .37 in 1999 to .77 in 2002. Hypertension has a much higher mortality rate than do stomach infections, but experiences a similar trend rising from 14.74 to 16.62 per 100,000.
In contrast to stomach conditions, heart disease is clearly the most fatal disease of the conditions considered. The age adjusted mortality of heart disease is always more than 150 per 100,000. However, the mortality rate from this disease has fallen dramatically over the period from 212 per 100,000 in 1996 to 163 in 2003.
Moreover, the mortality rate of this disease has not experienced a single year of increase over this period. Acute Myocardial Infarctions have also seen significant
decreases in mortality, falling from 71.8 per 100,000 in 1998 to 60.5 per 100,000 in 2002.
The suicide rate, a health outcome used to measure the efficacy of depression treatment, fell over the period 1996-2000 from 11.5 to 10.4. However, since 2000 the suicide rate has increased, if only slightly, from 10.4 to 10.8 per 100,000. The other health conditions examined have very flat trends over the period. Diabetes experiences a mild mortality rate increase over the period, whereas infant mortality falls slightly and skin cancer mortality remains exactly the same. The number of asthma attacks and the percent of people gaining mobility for arthritis also have not changed much over the period.
The values presented in Table 8 represent the average annual percentage change in treatment prices over the period. Price changes labelled as outcome in-corporate changes in health outcome measures, whereas price changes labelled naive do not account for health outcomes. Both the naive and the outcome-based mea-sures are determined relative to the earliest period for which the outcome measure is available. For instance, if the outcome-based measure is the age-adjusted mor-tality, then the price increase is defined as the percent increase of treatment prices needed to maintain the same level of age-adjusted mortality observed in the earli-est period - often the year 1996. The outcome-based measures are constructed by first determining the ratio of age-adjusted mortality rates in the given year to the age-adjusted mortality rate of the earliest year observed. For example, this ratio for diabetes mortality is 1.063 in 2003. This ratio is then multiplied by the naive price for the relevant year. The reported changes are the percentage change of this product over the period where outcomes are observed.
The importance of health outcomes are very important for determining the price of disease. For instance, some diseases, such as heart disease and acute my-ocardial infarction, have had improved health outcomes over the period, which has
Table 3.8: Price Increases for Outcome-Based Measures Dep- Heart
Region Statistic Diabetes ression Disease Arthritis Asthma
MSA Northeast Naive 0.039 0.032 0.064 0.042 0.046
Outcome 0.050 0.022 0.020 0.071 0.018
MSA Midwest Naive 0.057 0.044 0.042 0.040 0.046
Outcome 0.068 0.033 0.003 0.068 0.022
MSA Southeast Naive 0.050 0.037 0.046 0.037 0.047
Outcome 0.061 0.027 0.006 0.065 0.023
MSA West Naive 0.052 0.048 0.046 0.043 0.044
Outcome 0.063 0.037 0.006 0.072 0.018
Non-MSA Naive 0.049 0.042 0.037 0.033 0.056
Outcome 0.060 0.032 -0.001 0.060 0.029
Preg-Region Statistic AMI tension Infection Cancer nancy
MSA Northeast Naive 0.042 0.030 0.013 0.066 0.055
Outcome 0.004 0.066 0.297 0.066 0.029
MSA Midwest Naive 0.063 0.044 0.082 0.111 0.045
Outcome 0.021 0.082 0.440 0.111 0.016
MSA Southeast Naive 0.069 0.052 0.094 0.113 0.046
Outcome 0.026 0.090 0.467 0.113 0.037
MSA West Naive 0.050 0.018 0.055 0.049 0.045
Outcome 0.010 0.053 0.386 0.049 0.026
Non-MSA Naive 0.050 0.032 0.071 0.058 0.029
Outcome 0.011 0.068 0.418 0.058 0.024
mitigated much of the treatment price increases observed over the period. However, other conditions such as hypertension and arthritis have had deteriorating outcomes over the period, and thus accounting for health outcomes has exacerbated any price increases in those treatments. The importance of health outcomes is so severe that it can reverse the ranking of diseases. For instance, heart disease as measured by the naive estimate faces faster price increases in the Northeast than any other disease.
However, after accounting for health outcomes, heart disease treatment is second only to AMI as having one of the lowest rates of increase.
The price changes calculated using this method attributes the entire change in observed health outcomes to the quality of medical care services over time. At-tributing the entire fraction of observed changes to medical care productivity relies on strong assumptions about the demographic and behavioral compositional changes of society. For these results to be attributed to medical care, changes in diet, smok-ing habits, and environmental factors such as the safety of worksmok-ing environments must also not have changed.