SEARCH

SEARCH BY CITATION

Keywords:

  • SLE;
  • Medicaid;
  • Ethnicity;
  • Eligibility;
  • Costs

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. REFERENCES
  8. Appendix: APPENDIX A: ESTIMATED INSUER' COSTS

Objective

To investigate whether ethnicity was associated with differences in Medicaid eligibility, health care utilization, and direct medical costs in a systemic lupus erythematosus (SLE) population.

Methods

A retrospective analysis of California Medicaid claims data was conducted on patients with SLE. Patient eligibility and month-by-month utilization and costs were computed and compared across ethnic groups. Descriptive statistics are presented. A mixed regression model on patient-level data was used to verify the trends of the aggregate data, controlling for covariates. A survival regression model on time to ineligibility was used to show eligibility patterns adjusted for covariates.

Results

Hispanic patients were less likely to have a lengthy eligibility period as compared with other cohorts (∼50% versus 70% eligible at month 36, respectively). As treatment progressed, Hispanics generated lower total costs than other cohorts. Results for inpatient frequency, prescription costs, and outpatient/physician/supply (Part B) costs followed similar patterns. Mixed regression model findings revealed that when adjusted for age, sex, and aid program, total costs for Hispanic patients decreased as the length of care increased, in contrast to other ethnic groups. The interaction between ethnicity and treatment progression measured by quarter was significant (P < 0.0001), but ethnicity as a main effect was not (P = 0.091). This suggests that differences in total costs are small initially, but as the followup period extends, Hispanic patients experience lower total costs as compared with other ethnic groups.

Conclusion

These California Medicaid program data reinforce the importance of investigating treatment differences in ethnic groups with SLE.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. REFERENCES
  8. Appendix: APPENDIX A: ESTIMATED INSUER' COSTS

Systemic lupus erythematosus (SLE) is a recurrent, relapsing disease typically affecting young women and occurring in ∼0.1–0.2% of the population. Due to the recurrent nature of the clinical manifestations of SLE, patient health, quality of life, and associated costs of the disease vary with time, depending on disease status. Although the literature on the economic impact of this disease is somewhat sparse, studies suggest that SLE results in significant direct and indirect costs for the individual and society (1, 2).

The relationship between ethnicity and outcomes in SLE has gained considerable attention in the past decade. Minority populations have a higher overall prevalence of SLE and display poorer outcomes. Alarcón and colleagues (3) found that Hispanics accrued damage more rapidly (as measured by the Systemic Lupus International Collaborative Clinics Damage Index) than African American or white populations. Some recent studies suggest that differences in outcomes may be the result of socioeconomic differences associated with minority populations, as opposed to biologic or genetic differences between the races (4). Regardless of the specific determinants, it is evident that minority populations appear to be more negatively impacted by SLE than white populations.

Health care coverage also plays an important role in health outcomes, as research suggests that lack of coverage results in decreased access to care and poor health outcomes (5, 6). Between 1996 and 1997, the percentage of nonelderly Americans without health care coverage increased from 17.7% to 18.3%, and adults ages 18–64 account for the majority of this increase (7). Employment status and income play a major role in determining an individual's likelihood of having health insurance, and factors such as ethnicity, citizenship, and language barriers are closely related to employment and income.

Medicaid provides health care coverage to patients who qualify based on income level or medical need. Disenrollment in this program is a serious issue, and research suggests that of those who are terminated from the program, the majority remain uninsured (8). National data indicate that one-third of all Hispanics in the United States are uninsured, which is 3 times the rate of the non-Hispanic white population. Hispanics that are Medicaid eligible often become ineligible upon securing employment. However, they are more likely to work in occupations (i.e., agriculture) that do not provide health care coverage, and therefore are more likely to remain uninsured (9).

Based on the evidence suggesting poorer outcomes for SLE minority populations, it is critical to investigate the factors contributing to poor outcomes to intervene with a strategy for improvement. Evidence suggests that the Hispanic population is more likely to lack health insurance than non-Hispanic white persons (10). Lack of health insurance inevitably impacts access to primary and specialty care providers, and subsequently, patient health outcomes and associated costs. In the present study, we investigated whether ethnicity was associated with differences in Medicaid eligibility, health care utilization, and direct medical costs.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. REFERENCES
  8. Appendix: APPENDIX A: ESTIMATED INSUER' COSTS

A retrospective analysis of California Medicaid (known as Medi-Cal) claims data was conducted on patients with a diagnosis of SLE. Patient eligibility and month-by-month utilization and costs were computed and compared across ethnic groups. A mixed regression model on patient-level data was used to verify the trends in costs of the aggregate data, controlling for covariates. A survival regression model on time to ineligibility was used to show eligibility patterns adjusted for covariates.

Study population.

This study is based on a publicly available 20% sample of Medi-Cal Fee-For-Service claims data. The data include inpatient, outpatient, and prescription claims and eligibility information for all patients between January 1995 and June 2002. All Medicaid-managed care plans were excluded from this data set. For patients who were dually eligible for both Medicaid and Medicare, Medicare claims that were captured by the Medi-Cal crossover system were included in the data. Using an estimation procedure (see Appendix A), an attempt was made to recover the total payment made by Medicare for the crossover claims. Medicare payments that were not captured by the crossover system were unable to be estimated. The Institutional Review Board at the University of Southern California approved the study protocol.

Patients were selected for inclusion in the analysis if they met all of the following criteria: 1) a primary or secondary diagnosis of SLE (International Classification of Diseases, Ninth Revision code 710.0) during the claim period; 2) age 18 years or older on the index date (defined as the first date of SLE diagnosis in the claim period); and 3) eligible for at least 2 months prior to the index date in efforts to obtain new cases of SLE. Although a negative history (no SLE diagnosis) of 6 months would have been preferred, such a requirement resulted in losing about one-fifth of the patients (n = 469). To support our decision to use a criterion of 2 months negative history, we checked a sample of patients most likely to be new SLE patients, based on their 12 months' negative history and 12 months' eligibility for followup. One-third of these patients had at least 1 visit with a primary or secondary SLE diagnosis within 2 months of the index date, and 82% had at least 1 visit (non-SLE specific) within 2 months of the index date.

In total, 2,982 patients had a diagnosis of SLE between January 1995 and June 2002 and were ≥18 years of age on the index date. After qualifying them with 2 months of negative history, 2,395 patients remained for inclusion in the analysis.

Patient eligibility.

To compute patient eligibility, months relative to the month of the index date were used. The index month was month 1, and all utilization was measured subsequent to that month. For each of the relative months, the total number of patients who had been diagnosed with SLE and were not yet censored (due to the data end in June 2002) served as the denominator from which the proportion of eligible patients was calculated. To assess patient continuity in eligibility, we determined the total number of months between the first and the last eligible month and then calculated the proportion of eligible months. Time to ineligibility in the first eligibility phase was modeled by a survival regression adjusted for covariates.

Utilization and costs.

To track month-by-month utilization and costs per month eligible for each patient relative to the index date, the following method was applied.

Inpatient utilization and costs were credited based on the month of service; use of a nursing facility or an intermediate facility was credited to each month of actual utilization along with the associated costs; outpatient/physician/medical supply (abbreviated as Part B) utilization and costs were credited to the month of service; and prescription utilization and costs were credited to the month of prescription fill or refill.

Payments were used as a proxy measure for costs because this analysis was conducted based on an insurers' perspective. Total medical costs consisted of inpatient costs, Medicare Part B costs (as appropriate), prescription costs (Rx costs), and nursing facility/intermediate care facility costs (these were combined and abbreviated as LTC costs).

For crossover claims (claims that were eligible for both Medi-Cal and Medicare payment), Medicare covered most of the inpatient, Part B, and LTC costs and Medi-Cal made only the copayment and residual payments. Because some Medicare claims are not submitted to Medicaid, utilization and cost data for this subset of patients may be incomplete.

For inpatient services rendered by hospitals negotiated by the California Medical Assistance Commission, the payment information was suppressed for the recent 4 years because these contracts are considered proprietary. Please see Appendix A for details of the methods used to estimate these actual costs.

Inpatient costs, nursing facility costs, and intermediate care facility costs were each estimated with an average daily cost across the entire claim period of 8 years. Inflation adjustment to Rx and Part B costs was not applied because Medi-Cal payment to professionals was almost unchanged during the time period of the analysis, even though the billed amount was likely to change.

Statistical analyses.

The majority of the descriptive statistics are presented with figures detailing the activity during the monitoring period. For a given event (such as inpatient visits), either frequencies or average costs per patient-month from quarter to quarter are of interest. Each quarter is made up of 3 relative months. To compute the event frequency per patient-month for a given quarter, we first total the number of months when the event occurred across all patients in the quarter, then divide the sum by the total number of eligible months of all patients in the quarter. To compute the average costs per patient-month for a given quarter, we first sum the costs recorded in each month of the quarter, then divide the sum by the total number of eligible months of all patients in the quarter.

Smoothing methods.

A nearest-neighbor smoothing method is applied to tame large fluctuations that can result from a small sample size (11). This is done by taking a simple average of the numbers produced in the 5 quarters nearest to (and including) the quarter of interest. Typically, the 5 quarters would consist of the quarter of interest at the center with the other 2 quarters at each side, except when the quarter of interest is near the boundary of the monitoring period.

Because the number of eligible patients decrease as the followup period extends, the variation in utilization and costs can increase dramatically toward the end of the monitoring period. The time trends displayed in the fig- ures are based on aggregate data. To verify these trends, we used a mixed regression model on patient-level data controlling for certain covariates.

All analyses were performed using SAS for Windows, version 8 or version 9 (SAS Institute, Cary, NC) and Microsoft Excel ‘97 (Microsoft Corporation, Redmond, WA).

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. REFERENCES
  8. Appendix: APPENDIX A: ESTIMATED INSUER' COSTS

Patient demographics.

Patient demographic information (n = 2,395) is displayed in Table 1. Approximately 90% (n = 753) of the patients were women. Approximately 14% of the sample (n = 328) had missing or unknown ethnicity. About one-third of patients (n = 831) were classified in the age bracket of 30–44 years at the time of index diagnosis. Most of those in this age bracket were women (90.6%; n = 753).

Table 1. Patient demographics by eligibility type
 Regular no. (%)Dual no. (%)*Total no. (%)
  • *

    Dually eligible anytime during 1995–2002.

Age, years   
 18–29304 (12.7)51 (2.1)355 (14.8)
 30–44655 (27.3)176 (7.3)831 (34.7)
 45–54367 (15.3)161 (6.7)528 (22.0)
 55–64186 (7.8)167 (7.0)353 (14.7)
 ≥6521 (0.9)307 (12.8)328 (13.7)
Sex   
 Female1,401 (58.5)770 (32.2)2,171 (90.6)
 Male132 (5.5)92 (3.8)224 (9.4)
Ethnicity   
 White536 (22.4)358 (14.9)894 (37.3)
 Hispanic322 (13.4)49 (2.0)371 (15.5)
 African American279 (11.6)150 (6.3)429 (17.9)
 Others239 (10.0)134 (5.6)373 (15.6)
 Unknown157 (6.6)171 (7.1)328 (13.7)
Total1,533 (64.0)862 (36.0)2,395 (100.0)

Patients who qualified for either Medicare Part A or Part B coverage anytime during 1995–2002 were considered dually eligible. A majority of patients (93.6%) age ≥65 years on the index date were dually eligible. Hispanic patients were least likely to be dually eligible, with ∼90% of Hispanic patients <55 years of age as of the index date. About one-third (35.5%; n = 770) of the female cohort (n = 2,171) was dually eligible.

Eligibility results.

Because data are available through June 2002 only, a patient diagnosed with SLE in January 2002 did not contribute to the base of the seventh month or beyond. Thus, the number of patients contributing to each relative month decreased from 2,395 patients at month 1 to 117 at month 84. The data imply that ∼50% of patients remained eligible after 7 years (84 months).

At the end of year 3 (i.e., month 36), the proportion of eligible patients was 52.5% for Hispanics, 74.2% for whites, 78.2% for African Americans, 90.1% for others, and 70.4% for unknowns. These data illustrate that Hispanic patients were less likely to have a lengthy eligibility period, whereas patients classified in the “other” category (mainly Asians and Pacific Islanders) were more likely to have longer eligibility periods. The contrast in age between these 2 cohorts may be a contributing factor in explaining the difference, as Hispanic patients were considerably younger than the “other” cohort (P < 0.0001 on the 2-sample test of equal proportion with age <55 years), and therefore may have shorter eligibility periods.

The average eligible months from index diagnosis through June 2002 was 39.1 for whites, 31.2 for Hispanics, 42.9 for African Americans, 41.9 for others, and 36.0 for the unknown group. One way to measure continuity in eligibility is to compute a percentage of eligible months between the first and the last eligible month. This percentage was 87% for Hispanics and >95% for all other groups. These higher percentages suggest that continuity in eligibility was good across all groups.

Comparison of the type of aid program by ethnicity group was conducted. Hispanic patients admitted to Medi-Cal appear to be different from other ethnicity groups in terms of aid program composition. The Hispanic cohort was more likely to qualify based on medical need/indigent status (32% versus at most 22% for other ethnicity groups) or on the Omnibus Budget Reconciliation Act program (18% versus at most 1% for other ethnicity groups). More than 75% of patients in other cohorts qualified for Medi-Cal based on public assistance programs, whereas only 45% of patients in the Hispanic cohort qualified due to such programs (P < 0.001 on the test of equal proportion).

Utilization results.

Results of both inpatient and LTC frequencies per patient-month eligible data reveal trends showing Hispanic patients having the least inpatient utilization, whereas both Hispanic patients and those categorized as “others” generated the fewest LTC claims.

Costs of medical care.

Figure 1 displays a smoothed version of total costs per patient-month eligible for each quarter relative to the index month. The smoothed version is obtained by averaging the nearest 5 quarters at each point. In general, Hispanic patients generated the least total costs as compared with the other cohorts. The patterns for Rx costs and Part B costs are similar to those for total costs, with the Hispanic cohort again generating the least costs over time as compared with other cohorts.

thumbnail image

Figure 1. Total costs per patient-month smoothed with the nearest-neighbor method. To keep graphs simple, results for the “others” category were not included here, but were similar to findings for the white and African American groups.

Download figure to PowerPoint

Mixed regression model.

To confirm the time trend associated with Hispanic patients for total medical costs (shown with the aggregated data above), a mixed regression model was conducted based on quarterly patient-level data. The dependent variable of the model was the average total costs per patient-month eligible in a quarter. This was obtained by summing the total medical costs for each eligible month of the quarter, then dividing the sum by the number of eligible months during the quarter. This averaging operation reduces skewness in the cost data. Log transformation was not used because some averages were zero. An observation was considered missing for a quarter if the patient was ineligible that quarter. If the patient was eligible but incurred no costs, then the observed value was zero. The time trend to be confirmed was expressed as an interaction between quarter and ethnicity, which was adjusted for age, sex, aid type, and dual eligibility to Medicare and Medi-Cal. The patient effect was treated as a random effect because the quarterly observations for each patient may be correlated and patient specific.

The output of the model is shown in Table 2. The adjusted total costs for Hispanic patients decreased as the length of care got longer. The interaction between quarter and ethnicity was statistically significant (P < 0.0001), but the ethnicity as a main effect was not significant (P = 0.091, results not shown). This may suggest that the differences in total costs are small initially (near the index date), but as the followup period gets longer, Hispanic patients tend to experience lower total costs as compared with other ethnic groups, even when controlling for age and aid programs (Figure 2).

Table 2. Mixed regression model for adjusted total costs*
EffectEstimateSEPr > |t|
  • *

    OBRA = Omnibus Budget Reconciliation Act.

Intercept2,189.68546.5< 0.0001
Age, years   
 18–29−216.18178.80.2266
 30–44−490.96154.180.0015
 45–54−313.45158.320.0477
 55–64−302.36164.230.0656
 ≥650  
Sex   
 Female−142.7145.120.3254
 Male0  
Dual eligible   
 No−263.35102.920.0105
 Yes0  
Aid type   
 Assist−543.21513.850.2905
 Income−0.649652.950.9992
 Medical need−202.26522.040.6984
 OBRA431.84578.020.4550
 Others0  
Ethnicity   
 White−275.76138.850.0470
 African American−14.7492157.70.9255
 Hispanic−235.2178.040.1865
 Others−300.66162.390.0641
 Unknown0  
Quarter14.5914.28670.0007
Quarter* ethnic   
 White−0.25694.9130.9583
 African American−5.73635.37460.2858
 Hispanic−37.72655.9475< 0.0001
 Others−4.44185.54020.4227
 Unknown0  
thumbnail image

Figure 2. Adjusted total costs per patient-month based on a mixed regression model. To keep graphs simple, results for the “others” category were not included here, but were similar to findings for the white and African American groups. *Assuming that adjusted total costs = $1,000 for unknown group (not shown) at quarter 0.

Download figure to PowerPoint

Survival analysis.

A survival analysis on the length of the first eligibility period controlling for age, sex, and aid program difference showed significant differences in eligibility between Hispanics and other ethnicity groups (see Table 3). Survival analysis was applied to the length of the first eligibility phase. The log survival time in months was assumed to follow a Weibull distribution and was fitted with covariates including age, sex, dual eligibility, aid type, and ethnicity. Ethnicity remained significant (P < 0.0001) after adjusting for age, aid type, and other covariates in the model. Hispanic patients were more likely to experience a shorter first eligibility phase than patients of other cohorts.

Table 3. Survival model on log time in month of the first eligibility phase*
EffectEstimateSEP
  • *

    OBRA = Omnibus Budget Reconciliation Act.

Intercept3.70870.4852< 0.0001
Age, years   
 18–29−0.05200.15210.7323
 30–440.04290.13610.7529
 45–540.37800.14680.0100
 55–640.57970.16200.0003
 ≥650  
Sex   
 Female0.29530.12350.0168
 Male0  
Dual eligible   
 No−0.16940.09560.0764
 Yes0  
Aid type   
 Assist0.58550.45730.2005
 Income−0.39920.53680.4571
 Medical need−0.23280.46310.6152
 OBRA−0.14760.49060.7635
 Others0  
Ethnicity   
 White0.14940.11930.2106
 African American0.33520.13770.0149
 Hispanic−0.32920.14110.0196
 Others1.06280.1701< 0.0001
 Unknown0  
Scale1.1050.0313 
Weibull shape0.9050.0256 

Sensitivity tests.

With the 6-month negative history requirement, 469 patients were excluded and 1,926 patients remained. The same mixed model was fitted on the average quarterly patient-level data for total costs to test the sensitivity of results. The interaction term between quarter and ethnicity was still statistically significant (P < 0.0001), suggesting a decreasing trend in total costs for Hispanics as treatment progressed. Although the magnitude of estimates changed, the overall trend for ethnicity remained the same.

The second test was conducted after exclusion of all dually eligible patients, because not all Medicare costs for this population were recoverable. After exclusion, 1,533 patients remained. The same mixed model was fitted on the average quarterly patient-level data for total costs. The interaction term between quarter and ethnicity was still statistically significant (P < 0.0001), suggesting a decreasing trend in total costs for Hispanics as treatment progressed. The magnitude of estimates related to Hispanic and quarter changed only slightly, and the overall trend for ethnicity remained the same.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. REFERENCES
  8. Appendix: APPENDIX A: ESTIMATED INSUER' COSTS

In this study, we explored eligibility patterns, utilization trends, and direct costs in a Medi-Cal lupus population. In general, Hispanic patients were less likely to have lengthy eligibility periods, whereas patients placed in the “other” category (primarily Asians and Pacific Islanders) were more likely to have longer eligibility periods. A partial explanation may be due to the Hispanic population in this sample being considerably younger than the other cohorts: 95% of individuals aged ≥65 years were insured through Medicare, but individuals <65 years were more likely to have multiple forms of insurance or no insurance at all (12). Younger individuals may need Medi-Cal assistance for only a short period of time until employment is secured, providing private insurance and/or raising the income level so that the patient no longer qualifies for aid.

Further investigation into differences in eligibility patterns suggests that Hispanic patients qualifying for Medi-Cal are different from other ethnicity groups in terms of aid program composition. They are more likely to qualify based on medical need/indigent status or under the OBRA program. Certain aid programs, such as OBRA, have restrictions on the length of eligibility. Although all of these factors contribute to shorter eligibility periods in Hispanics, they are far from complete. A survival analysis on the length of the first eligibility period controlling for age, sex, and aid program difference still showed significant differences in eligibility between Hispanics and other ethnicity groups.

Hispanic patients had the least utilization of Medi-Cal care services and generated lower total costs as the care of SLE progressed. Again, the younger age of the Hispanic cohort may have potentially contributed to this trend because younger patients tend to recover faster, are less likely to have major comorbid disorders, and treatment is generally less expensive due to fewer complications. Previous studies suggest an association between older age and greater damage accrual, most likely due to other comorbid disorders related to aging (13). However, one prior study revealed that older age was associated with lower costs, possibly due to reduction in SLE disease activity as patients age (especially those postmenopausal) (1). An analysis of total costs with longitudinal patient-level data was attempted using mixed-effect models to adjust for age, sex, dual eligibility, and aid program difference. However, the utilization trend for Hispanics still prevails after the adjustment. Although age and program type have demonstrated some contribution to these trends, further investigation is necessary to determine other influential factors.

Hispanic patients also generated significantly lower prescription drug costs than the other ethnicity groups. One explanation may be that the Hispanic cohort is receiving less optimal care than other cohorts. For example, the care providers may not be prescribing the newer (and possibly more effective) treatments that tend to be more costly (i.e., newer immunosuppressive agents). These findings may also be explained by the Hispanic cohort being younger and less likely to have other major comorbid disorders, resulting in lower prescription costs than other cohorts. Another explanation is that the Hispanic population may not be filling their prescriptions at the same rate as the other cohorts, and therefore generating less cost. The reasons for Hispanic differences in prescription drug utilization may prove to be critical in understanding the impact of differences in outcome for minority populations in this disease in which pharmacologic management is thought to be quite important.

This study is limited in that it utilizes Medi-Cal claims data with no access to clinical measures or patient-reported health outcomes. Therefore, we were unable to determine how the eligibility, utilization, and cost trends in this study relate to clinically important patient outcomes. Additionally, ethnicity information in administrative data may be flawed. Self-report information on ethnicity has only recently been added to Medicare files, and is still in the process of being updated. In the past, this race indicator has been somewhat unstable. There are also some limitations related to the mixed regression model applied in this study. The quarterly number of average total costs may not be normally distributed, and the variance structure of the random effect may be more complicated. Lastly, because Medicare does not consistently submit claims to Medi-Cal for the dually eligible population, we can not be sure we have complete claims history for this subset of patients. However, access to this missing data would only bolster our argument, as we would expect utilization and costs to further increase for this population, hence widening the gap between Hispanics and all other groups. A sensitivity test reveals the same pattern for patients who were not dually eligible and thus were less likely to be older than age 65.

Because SLE affects young women of reproductive age, assessing indirect costs is critical to understanding the impact of this disease. Sutcliffe and colleagues (1) found that direct costs accounted for one-third of the total costs for SLE patients in the UK, whereas the other two-thirds of the total costs were attributable to indirect costs. This highlights the importance of assessing indirect costs when investigating the economic and societal impact of SLE.

These preliminary data in a single state Medicaid program reinforce the importance of ethnicity in health care, particularly in lupus. Further studies are needed with additional complementary information concerning indirect costs. Furthermore, if it is true as previously suggested that Hispanic lupus patients accumulate damage faster than other minority groups (3), the fact that these patients appear to generate lower total direct costs suggests that the cause of this phenomenon needs to be explored further. The finding that Hispanic lupus patients have shorter eligibility periods than other cohorts should prompt further investigation into the reasons this population is likely to lose Medicaid fee-for-service eligibility, transition to Medicaid managed care, or obtain alternative health care coverage.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. REFERENCES
  8. Appendix: APPENDIX A: ESTIMATED INSUER' COSTS
  • 1
    Sutcliffe N, Clarke AE, Taylor R, Frost C, Isenberg DA. Total costs and predictors of costs in patients with systemic lupus erythematosus. Rheumatology (Oxford) 2001; 40: 3747.
  • 2
    Gironimi G, Clarke AE, Hamilton VH, Danoff DS, Bloch DA, Fries JF, et al. Why health care costs more in the US: comparing health care expenditures between systemic lupus erythematosus patients in Stanford and Montreal. Arthritis Rheum 1996; 39: 97987.
  • 3
    Alarcon GS, McGwin G Jr, Bartolucci AA, Roseman J, Lisse J, Fessler BJ, et al. Systemic lupus erythematosus in three ethnic groups. IX. Differences in damage accrual. Arthritis Rheum 2001; 44: 2797806.
  • 4
    Karlson EW, Daltroy LH, Lew RA, Wright EA, Partridge AJ, Fossel AH, et al. The relationship of socioeconomic status, race, and modifiable risk factors to outcomes in patients with systemic lupus erythematosus. Arthritis Rheum 1997; 40: 4756.
  • 5
    Weissman JS, Witzburg R, Linov P, Campbell EG. Termination from Medicaid: how does it affect access, continuity of care, and willingness to purchase insurance? J Health Care Poor Underserved 1999; 10: 12237.
  • 6
    American College of Physicians- American Society of Internal Medicine. No health insurance? It's enough to make you sick. Philadelphia: American College of Physicians–American Society of Internal Medicine; 2000.
  • 7
    Fronstin P. Sources of health insurance and characteristics of the uninsured: analysis of the March 1999 Current Population Survey. Washington (DC): Employee Benefit Research Institute; 2000: 126.
  • 8
    Short PF, Freedman VA. Single women and the dynamics of Medicaid. Health Serv Res 1998; 33: 130936.
  • 9
    Quinn, K. Working without benefits: the health insurance crisis confronting Hispanic Americans. New York: The Commonwealth Fund, Task Force on the Future of Health Insurance for Working Americans; 2000. Publication 370. URL: http://www.cmwf.org/.
  • 10
    US Census Bureau. Health insurance coverage. URL: http://www.census.gov/hhes/www/hlthin01.html.
  • 11
    Altman NS. An introduction to kernel and nearest-neighbor nonparametric regression. Am Stat 1992; 46: 17585.
  • 12
    Powell-Griner E, Bolen J, Bland S. Health care coverage and use of preventive services among the near elderly in the United States. Am J Public Health 1999; 89: 88286.
  • 13
    Alarcon GS, Friedman AW, Straaton KV, Moulds JM, Lisse J, Bastian HM, et al. Systemic lupus erythematosus in three ethnic groups. III. A comparison of characteristics early in the natural history of the LUMINA cohort. Lupus 1999; 8: 197209.

Appendix: APPENDIX A: ESTIMATED INSUER' COSTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. REFERENCES
  8. Appendix: APPENDIX A: ESTIMATED INSUER' COSTS

The following methods were used to estimate insurers’ costs. This was necessary because actual costs for inpatient services rendered by California Medical Assistance Commission hospitals during the recent 4 years and actual costs of crossover claims for Medicare are unknown.

2) Inpatient costs were derived by multiplying the number of service days by the average daily cost estimate of $1,000 for all patients.

3) Nursing facility costs were derived by multiplying the number of service days by the average daily cost estimate of $110 for all patients.

4) Intermediate care facility costs were derived by multi- plying the number of service days by the average daily cost estimate of $140 for all patients.

5) Part B costs were estimated by 4 × Medicare coinsurance + Medi-Cal paid amount + paid amount of other parties for crossover claims. For noncrossover claims, Part B costs are the sum of the amount paid by Medi-Cal and other parties.

6) Prescription costs included payment made by Medi-Cal and other parties.