Technical Supplement. The cost of health insurance administration in California: Insurer, physician, and hospital estimates

12  Download (0)

Full text

(1)

Editor’s Note: This online data supplement contains supplemental material that was not included with the published article by James G. Kahn and colleagues, “The Cost of Health Insurance Administration in California: Estimates for Insurers, Physicians, and Hospitals,” Health Affairs 24, no. 6 (2005), online at

content.healthaffairs.org/cgi/content/abstract/24/6/1629.

Technical Supplement

The cost of health insurance administration in California:

Insurer, physician, and hospital estimates

We estimated the percent of costs that is billing or insurance-related in three health care settings: private insurers, physician offices, and hospitals. To accomplish this, we abstracted administrative cost elements from existing data sets, calculated overhead attributable to these costs, and estimated the percent of each cost element that is billing or insurance-related (BIR). Within individual settings, we compared BIR to the best

estimate of total revenue in the relevant data sets.

Below we describe the data sources, cost categories, and estimation of BIR for each of the three settings. We also describe how we estimated total percent BIR spending in California acute health care funded through private insurance. At the end of this technical supplement we provide a discussion of our method of BIR assignment.

Our analysis best reflects administrative costs in 2000. Our data sources clustered around that year. It was impossible to formally age the key input data (i.e., % of costs attributable to administration) since economic inflation indices document prices and costs

(2)

rather than functional distributions. In addition, the insurer data were available to us for 5 years in the aggregate rather than for individual years.

Private insurers

Data sources: We used data collected by a consulting company, Milliman USA,

because it provided a large and varied sample. Data were collected by Milliman through on-site evaluations at 73 insurers (49 for-profit and 24 not-for-profit). These insurers operated 129 insurance plans (63 commercial, 43 Medicare, and 23 Medicaid),

representing a range of reimbursement mechanisms. Seven insurers were in California, representing 5 commercial, 3 Medicare, and 5 Medicaid plans. Membership in the plans outside California averaged 782,000 (median 257,000); in California, 2.1 million and 250,000 respectively. Data were collected from 1996-2001, 15-27 plans per year. Costs were recorded in nominal dollars; Milliman reports that over the time period there were both improvements in efficiency and partially offsetting wage inflation. Costs in

commercial plans were defined as referent; costs for Medicaid and Medicare plans were calculated based on observed differences in intensity of services per member per month, by functional category. The data were provided to us in the aggregate, as median cost per member per month for each plan type and cost element. Milliman estimated the median commercial premium per member per month, used as the denominator in calculating percent BIR, as $177, with values of $153 and $650 for Medicaid and Medicare, respectively.

(3)

Cost categories: Costs in the Milliman data are divided into 13 categories: general

administration, claims (including preparation, data capture, adjudication, coordination of benefits, adjustment, and auditing), sales and marketing, finance and underwriting, membership and billing, provider services and credentialing, customer service,

information systems, utilization and quality review, case management, medical director, and other healthcare services. In each category, Milliman collected data on the salary costs of employees engaged in the activity. Milliman estimates that for every $1.00 of salary there is an associated $1.50 in overhead, representing the costs of employee

benefits, information technology, facilities, and other overhead. External brokers fees and profits are excluded.

Estimation of BIR: We assigned BIR based on descriptions of each cost category,

and consensus of the first two authors. These are summarized in Table 1. For categories that appear to focus entirely on reimbursement (e.g., claims), we assume 100% BIR. For categories that are likely to be predominantly about eligibility or reimbursement (e.g., customer services), we assume a high percent (e.g., 85%). For categories with mixed clinical and insurance functions (e.g., case management), we assume about half. Finally, for one category with a large clinical focus (e.g., utilization & quality review), we assume only 25% BIR.

Physician offices

Data sources: For total cost by category, we used data from a survey mailed

(4)

practices. We analyzed 2000 data (from the survey completed in 2001), for the western region (Alaska, Arizona, California, Colorado, Hawaii, Idaho, Montana, Nevada, New Mexico, Oregon, Utah, Washington, and Wyoming). We used the western region rather than California alone due to sample size considerations; the findings were validated by the California medical group interviews described below. The valid response rate was 23%. We used responses organized into 3 categories: multi-specialty (52 groups, with 2,682 physicians), single specialty primary care (27 groups, 221 physicians), and single specialty surgery (15 groups, 84 physicians). Means are weighted by number of

physicians. Physician time spent on administration, by function, is estimated from a national survey conducted by the American Medical Association.i

Cost categories: Administrative costs in the MGMA survey are divided into 13

categories: general administration, other administrative support, contracted services, administrative supplies and services, business office, medical reception, managed care administration, information technology, MSO fees, medical secretarial, transcription, medical records, and provider work on administration. Overhead proportionally allocated to these categories includes benefits for non-provider staff; housekeeping, maintenance, and security; building and occupancy; furniture and equipment; and other.

Estimation of BIR: We assigned BIR by cost category based on data we collected

from eleven California medical groups (five multi-specialty, three single specialty

primary care, and three surgery). We choose the practices based on a convenience sample of 12 offices representing a mix of specialty, size, and geographic location. We recruited each practice (11 of 12 agreed; the refusal was due to unavailability of a needed

(5)

Using a structured interview, we asked how the practice performs the work in each MGMA category (e.g., “Medical Reception”). Specifically, we asked for the number of people working in the category and their salary and benefits, and, for each person, the proportion of time spent on BIR versus other administrative activities. We provided examples of BIR and non-BIR tasks. The respondent often consulted other staff or records, and sometimes brought in another respondent. During pretesting, some

respondents found it difficult to divide functions according to the MGMA categories. To address this problem, we disaggregated some MGMA categories into activities familiar to group practice administrators. For example, we divided the “business office” category in MGMA into “front office billing” (collecting information required to submit

reimbursement claims) and “back office billing” (submitting the claim and tracking payment), which correspond to distinctions commonly made by our respondents.

The distribution of administrative costs was similar in our practice interviews and in MGMA data. Mean total administration costs for work done by non-provider staff was 20.7% in our sample (weighted by practice expenses) and 21.0% for MGMA (weighted by number of physicians). For four categories, the percentage of total costs in our 11 practices was within 0.1% of the percentage in the MGMA data (business office,

information services, medical records, medical transcription) and for four categories the differences were greater (general administration 1.2%, managed care administration 0.4%, medical reception 1.8%, contract staff 1.3%).

Across the 11 practices, the BIR percent for each administrative cost category (reported in Table 2) was most consistent for categories contributing the most to BIR. A standardized measure of variation (standard deviation divided by the mean) was lowest

(6)

for the four categories representing 70% of non-provider BIR costs: 0.29 for business office, 0.25 for managed care administration, 0.41 for information services, and 0.70 for medical reception. This measure of variation was higher (0.98 to 2.02) for the four categories representing 30% of non-provider BIR costs. This suggests consistency across practices in our respondents’ judgments about the percent of activity devoted to BIR in the areas responsible for most of BIR activity, but more variation in activities accounting for smaller parts of BIR.

A substantial fraction of BIR activities in physician offices is provided by physicians and other clinical providers. We did not ask systematically about physician BIR activities in our survey of 11 practices, but relied instead on a collateral source. The AMA survey of physicians found that physicians spend 5 hours per week (8% of their time) on administration and other non-clinical tasks including billing, handling claims, managing a practice, medical staff functions, and teaching.i Based on the breakdown of non-clinical tasks ii and our estimates of BIR percent, we estimated that 4.9% of

physician time is dedicated to BIR, representing 2.1% of practice revenue. In addition, in our survey of practices we found that non-physician clinical staff contribute 7.5% of their time to BIR, representing 1.6% of practice revenue.

Hospitals

Data sources: We used hospital financial reporting data collected by the

California Office of Statehood Health Planning and Development (OSHPD). All

(7)

Annual Financial Disclosure Reporting Form. We used data for fiscal year 1999

(OSHPD year 24). We removed psychiatric, long-term care, and rehabilitation hospitals from the data base prior to analysis. The remaining acute general and children’s hospitals numbered 392. Almost all hospitals in the state reported data for each of the required data elements. The significant exception was Kaiser Foundation hospitals, which omitted many categories of information, especially for Southern California; we were able to include Northern California hospitals in broad category totals (i.e., administrative and fiscal) and a subset of specific categories.

Cost categories: Administrative costs are divided into 21 categories. Nearly half

of costs are reported to be in three broad categories: hospital administration, other administration, and general accounting. Other relatively large categories are patient accounting, credit and collections, public relations, personnel, nursing administration, admitting, and medical records. Overhead proportionally allocated to these categories includes employee benefits, non-patient food, purchasing and stores, communications, printing/duplicating, insurance other, other unassigned, and net operating revenue.

Estimation of BIR: We assigned BIR based on descriptions of each cost category,

and consensus of the first two authors. These are reported in Table 3. For categories that focus entirely on reimbursement (e.g., credit and collections and patient accounts), we assume 100% BIR. For categories that are likely to be predominantly about

reimbursement (e.g., utilization management), we assume a high percent (75%). For categories with substantial but not exclusive BIR functions (e.g., admitting), we assume 50%. For functions with clinical functions (e.g., medical staff), we conservatively assume

(8)

0%. Finally, for two large general administrative categories (e.g., hospital administration and other administration), we used a range of 25% to 75%.

Standardization across health care setting

To facilitate comparison across setting, we grouped the spending into 4

categories: Broad administrative represents nonspecific administration categories; these are often relatively large. The estimated proportion due to BIR varies by setting. Claims

billing or payment represents activities explicitly linked to movement of funds related to

insurance. The BIR proportion is very high. Other specific administrative represents explicitly defined administrative categories. The proportion due to BIR is typically low to moderate in health care delivery settings, and high in insurance settings. Major clinical

element represents activities with clear clinical components. The proportion due to BIR

varies widely.

Total across settings

We also estimated the contribution of BIR administrative costs to acute medical care services (hospital and physician) funded through private insurance in California. To create a total estimate of BIR across settings, we highlighted the BIR incurred for

privately insured care since this is the payer subset for which we had the best insurer BIR data (e.g., we did not examine BIR for public agencies administering Medicaid). Thus, we used the commercial BIR estimate for insurers, and for providers we assumed that BIR as a percent of revenue is the same for privately and publicly-insured care.

(9)

We used the BIR values for commercial plans and multi-specialty groups, and the range for hospitals. The total percent BIR for services paid through this mechanism is calculated with the following formula (defined and explained below):

TBIR = IBIR + (1 – IADM) * [H%$*HBIR +(1-H%$)*PBIR]

Where: IBIR = percent BIR for insurers, including a profit allocation

IADM = the percent of private insurance premiums retained for

administrative activities (including BIR activities, other activities, and profit)

H%$ = the percent of medical services payments made to hospitals

HBIR = percent BIR for hospitals

P%$ = (1-H%$) = the percent of medical services payments made to

physicians (in this calculation we consider only hospital and physician payments).

PBIR = percent BIR for physicians

In calculating IBIR, we allocate insurer profits (or, for not-for-profit plans,

operating surpluses) proportionally to BIR and non-BIR administration, since reductions in administrative functions should similarly reduce operating surpluses. Profits or operating surpluses are estimated from two sources. Mandatory Knox-Keene reporting indicates pre-tax profits of 3.6% for FY 2000-2001 (mean, weighted by enrollment), which has been relatively stable over the last five years.iii Another study reports pre-tax profits of 3.4% for 2000, and found that investment income equaled 20% of profits.iv We use 2.72% to exclude investment income, which does not increase BIR costs. Further, we report the result using post-tax profits of 2.3% (adjusted for investment income = 1.84%),

(10)

on the theory that profits net of tax may better represent true costs to society for BIR activities. We also report the result excluding profits, for similar reasons.

The factor (1 – IADM) assures that the BIR proportions for hospitals and physicians

are applied only to the amount paid for health care. The terms H%$ and P%$ are the

portions of acute medical care spending representing by each type of provider. These are set at 55% and 45%, based on estimated 2001 private California spending for these providers of $18 and $15 billion, respectively, derived from the National Health Accountsv limited to appropriate providers and services.vi HBIR and PBIR are the BIR

percents reported above.

To simplify the analysis, we restrict our estimates of BIR to insurers, hospitals, and physicians. Although a more complete analysis would include estimated BIR for pharmacy and other types of insured services, our analysis captures the major types of acute care spending, and 80% of all privately insured health care spending.

Comment on BIR assignment

The two first authors estimated the BIR percents; the other two authors did not assist with BIR assignment. As noted in the main article’s Discussion, the authors’ biases about the extent of BIR are countervailing, so the consensus reflected considerable discussion and often the use of mid-range values. Still, we note that some BIR

assignments are imprecise, in particular those for categories with mixed functions. We believe that some imprecision is inevitable for several reasons. First, for those

administrative functions that serve mixed (BIR and non-BIR) purposes, the relative effort to each purpose is not reported. Second, even with facility-level detail, there would still be judgment calls about what purposes are served by administrative tasks.

(11)

Nonetheless we were encouraged on the robustness of BIR by two considerations. First, many administrative tasks can be reasonably assigned 100% or 0% BIR – they are, by definition, entirely or not at all for BIR. Second, when we compared the BIR levels reported across our sample of 11 physician offices, the BIR percents assigned for categories with large BIR components varied little across practices.

A few examples are illustrative.

We assigned 100% of insurers’ “sales and marketing” to BIR. This is because competition for customers among insurance plans is entirely dedicated to insurance purposes; no clinical or other administrative function is served that we could determine. How much of the BIR could be avoided with alternate insurance mechanisms depends on the need for continued sales and marketing. How insurance product competition affects the cost of health care is another important, but poorly understood, consideration (see main paper Discussion).

We assigned 0% BIR to hospitals’ 0.6% “medical staff” administration costs. We believe that this assignment is conservative, in that some medical staff credentialing may be driven by insurer requirements. Nonetheless, we expect that most medical staff activity focuses on maximizing the quality of the physicians working at the hospital.

Utilization management in hospitals (assigned BIR 75%) is a challenging example. Much of this activity is devoted to complying with insurer rules regarding lengths of stay, although some of this activity has, as its main purpose, the goal of allowing the hospital to manage within the fixed budget provided by capitated payers or from Medicare DRGs. The portion of UM that is devoted to complying with insurer rules

(12)

we treat as BIR; the portion that is simply part of the hospital management efforts we consider to be non-BIR. We recognize that others may estimate a different percent BIR, and thus present adequate detail to allow them to do so. If BIR for utilization

management were 25%, this would decrease the total BIR for hospitals by 0.5% of total revenue. Further, by assigning a BIR of 75%, we make no assumption regarding the portion of the BIR that could be eliminated with alternate insurance mechanisms.

i

Zhang P, Thran SL, eds. Physician socioeconomic statistics, 1999-2000. Chicago, IL: Center for Health Policy Research, American Medical Association, 1999.

ii

Cost and coverage analysis of nine proposals to expand health insurance coverage in California: Final report. Prepared by The Lewin Group, Inc. for the California Health and Human Services (CHHS) Agency, March 31, 2002.

iii

California Medical Association. 9th Annual Knox-Keene health plan expenditures summary, FY 2000-01. February 2002.

iv

Baumgarten A. California Managed Care Review 2002, prepared for California HealthCare Foundation, 2002

v

Levit K, Smith C, Cowan C, Lazenby H, Sensenig A, Catlin A. Trends in U.S. health care spending, 2001. Health Aff (Millwood) 2003;22(1):154-64.

vi

Selden TM, Levit KR, Cohen JW, et al. Reconciling medical expenditure estimates from the MEPS and the NHA, 1996. Health Care Financ Rev 2001;23(1):161-78.

Figure

Updating...

References

Related subjects :