Hospital-based radiologists are the single largest submitter of insurance claims of any specialty in the country. Annually, the processing costs to the radiology providers and insurance carriers are more than $1.5 billion. This figure does not factor in the claims pertaining to cases done outside the hospital departments. The costs of the receivable system represent 35% to 50% of the practice operating expenses. These expenses exclude physician benefits and malpractice insurance.

The radiology practice covering a hospital department will create 25 to 50 times the number of new patient accounts, per year, as the busiest referring physician group. Because of the extraordinary account volumes and reliance on hospital information, the only way to cost-effectively handle the processing is through technologically advanced computer platforms. It is not economically possible to hire enough personnel to process the information by hand. There is still a need for manual intervention, but the most sophisticated systems will perform as much of the processing functions as possible.

It is unfortunate that this complexity and cost are required because, to a certain extent, both the radiology practice and the hospital are processing similar information pertaining to these cases. It is not likely that alternative methods will be forthcoming because radiologists have no interest in alternatives.


All accounts receivable systems are structured to handle the submission of claims to insurance carriers where it is expected that the carrier will respond to a charge for each procedure on the claim. This is the predominant method of payment in the radiology marketplace. The primary alternative is capitation, where the radiology practice receives a fixed amount per month based on the number of patient lives they are under contract to manage. The discussions in this article will deal only with how receivable systems handle fee-for-service reimbursement.

A typical radiology practice will submit insurance claims to between 150 and 300 carrier populations. The payment characteristics of these carriers will fall into the categories listed in Table 1. We have set this universe to consist of 10 procedures, each one billed to a separate carrier category at a fee of $100. This will facilitate the illustration of collection ratios. This would be considered nonweighted population.

We will start by explaining the way the Medicare program compensates radiologists (line 01). This federal carrier is the largest payor for radiology services in the country. Its policies often influence the mechanics for payment by other major carriers in the marketplace. The radiology practice participates in the Medicare program, meaning that it will accept the Medicare fee as payment in full. Column B illustrates the payor allowance for each insurance carrier; the Medicare allowance has been set at $28. This fee is proportionately realistic when considering the relationship between average radiology fees and the 1999 Medicare rates.

Participation requires the radiology provider to write off the difference between the practice fee and allowance: essentially, someone must go into the system and post a noncash contract adjustment. This is illustrated in column C. Also, to keep the model simplified, we are not considering Medicare’s annual assessment of a deductible. Medicare regulations stipulate that the program will pay a participating provider 80% of the allowance, requiring the beneficiary to pay 20% (coinsurance). Column E represents the cash payment from the carrier, column F illustrates the amount that must be billed to the patient.

Whenever a patient is responsible for some or all of the charges, there will be a bad-debt risk. This risk can be high for Medicare patients because they often have difficulty understanding their obligations or have limited resources. Conversely, many Medicare patients purchase supplemental insurance that covers annual deductible and coinsurance requirements. In this model we have arbitrarily set the bad-debt risk at 30%. Column H shows the computation of the bad-debt amount on the patient balance for each carrier. Column I completes the calculation by combining the payments from the carrier and the patient. This number represents the cash value of the charges attributed to each carrier in this patient population.

The next carrier is the state’s Medicaid program, supported by state tax dollars and federal subsidies. The fee schedule is typically more conservative than Medicare, but payment is made in full. There can be potential bad-debt risk when expenditures exceed state budgets near the end of the fiscal year. Supplemental funding is often not used to respond to these types of claims.

Workers’ compensation and personal injury coverage have one thing in common. They both can involve litigation, which can delay substantially payment of the claim. Most states set the rates for workers’ compensation, although employers purchase coverage from private carriers. The fees paid under personal injury coverage tend to be more generous. This coverage also pertains to automobile accidents.

The Blue Shield and managed care plans have similar payment characteristics. There has been a recent tendency for these plans to set their fees as multiples of Medicare rates. The plans with more conservative fees will make direct payment to the radiology provider. Those with higher rates will require the patient to meet a coinsurance obligation.

The indemnity contracts are phasing out of the marketplace because the premiums are prohibitively expensive. Indemnity contracts are defined as those that allow the radiology provider full recourse for its total fee. Some carriers will pay the full fee directly to the radiology provider; others will require the patient to meet a coinsurance obligation.

Self-pay patients are generally indigent and secure most of their medical services in the emergency department. Large proportions of these charges are written off as bad debt.

Line 11 shows the total of each column. The gross charges are $1,000. However, because the radiology provider has agreements with various carriers, the maximum that can be collected on this nonweighted universe is the total in column B, $590.05. The provider is obligated to adjust the charges by the difference between the fee and allowance, the total of column C, $409.95. The total of column I combines payments from the carrier and patient, net of the bad-debt risk.

Line 12 shows the gross collection ratio, the relationship between actual cash and original gross charges. This used to be the standard benchmark for measuring collection efficiency. However, as carrier initiatives have driven fee reimbursement down, it has affected the validity of this benchmark. Why? If a major carrier successfully negotiates a reduction in price, the future gross collection ratio will go down even if the efficiency of the system remains the same.

The more appropriate benchmark is one that measures actual cash collections against the maximum cash value of the charges. This is referred to as adjusted gross charges, defined as gross charges minus contract adjustments. Line 14 shows the net collection ratio. Even if the carriers in this universe are modifying fees to the radiology provider, there will be little or no impact on the net collection ratio because the contract adjustments would be removed from the denominator. The only factor that would affect the net collection ratio is bad debt, which does reflect the efficiency of the receivable system.

The last two indices are the gross and net bad-debt write-offs. We also suggest that the bad-debt write-offs be expressed as a percentage of adjusted gross charges for the same reasons given for the collection ratio. Why measure bad debts against a charge base that cannot be collected in full? Also, using net ratios provides a full explanation of the overall efficiency of the system to liquidate a finite receivable universe. When the receivable balance on this universe reaches zero, the sum of the net collection and bad-debt ratios will be 100%.


We will now draw from the lessons learned in the review of Table 1 by examining the implications of structural differences in patient populations. The box at top left in Table 2 illustrates the third-party populations of Practices A, B, C, and D. For example, the average Medicare population of Practice A is 45% of the total. The next largest population is the grouping of managed care plans that pay at a fee schedule equal to 200% of the Medicare.

The analysis at the bottom of the table draws from the reimbursement characteristics of Table 1, but now considers the material differences in the third parties that pay the claims. We will illustrate the mathematics of this table by walking you through the calculation of the cash revenue of practice A attributed to its Medicare patients (line 01A, column A1). The population still consists of 10 patients of which Medicare covers 4.5. It pays $26.32 per examination multiplied by 4.5 patients, equaling $118.44. The same mathematics applies to the computation of the contract adjustment and bad-debt losses for each carrier within each practice universe.

Lines 14A to 17A are the ratio benchmarks for each practice. We will consider that each practice’s receivable system is equally efficient. Each one secures maximum cash value from its universe of charges. The different results are accounted for by the way each payor has negotiated reimbursement with the practice. There is no such thing as a standard collection ratio that applies to practices in a specific city or region.

Many readers will consider the mathematics of Table 2 to be obvious, yet there are radiologists who become alarmed upon learning that their counterparts covering Hospital B across town are collecting a higher percentage of charges. Perhaps there is a problem, but the easiest first step is to first find out how different the patient populations are before going to steps 2 through 4.


To appreciate the effort required to achieve maximum efficiency levels, we will portray a database extraction that shows collection results for a specific month-of-service. The information is a real-life example of how a billing system captured and billed clinical charges and worked toward the adjudication of the accounts. Few practitioners understand how long it takes to completely resolve a finite universe of charges. Most standard system reporting packages do not illustrate information in the manner we are going to present. This information is available because of the database features of this specific system.

Table 3 is structured to show the continuous resolution of charges attributed to month-of-service referred to as Month 1. Lines 01 to 06 show how receivables are reconciled by considering that new charges increase the balance, and cash, contract adjustments, and bad-debt write-offs reduce it. Line 02 also shows that not all of the charges attributed to the month-of-service are captured and billed in that month.

Advanced receivable systems electronically interface with the hospital information systems. Interface programs are written to extract information fields. These fields have been organized in specially constructed hospital files containing demographic and clinical information from each admission that is relevant to the processing needs of the radiology receivable system. Some practices also continued to receive a hard copy of the actual examination interpretation, using this as a basis for entering examination codes. This principally applies to the expensive surgical or interventional cases.

The specialty files are often consolidated either when a discharge date has been entered in the patient’s hospital file or when the institution creates its own claim to the insurer. This is generally within 24 to 72 hours for outpatient cases. Inpatient stays may range between 4 to 8 days and may cross over to a subsequent month. Also, the hospital must wait for medical records to validate the discharge diagnosis before it can submit claims. This may delay the release of claims for up to 72 hours after discharge.

Because the account volume is so large, even for a medium-size practice, it is wise to consider waiting until the hospital constructs its own claim file before performing an interface extraction. This guarantees that the radiology receivable system will receive the most accurate information it is going to get from the hospital.

Most good systems will have edit programs that test the validity of the information fields. It will identify accounts and fields that require additional work by the billing staff; sometimes the information needs of the radiology system differ from the hospital’s and certain fields may contain incomplete or missing data. This process will further delay the creation of an account.

In the model illustrated here a charge will appear only when the system can submit a complete claim. The charges are held in suspense file until the staff captures relevant field information. The majority of charges were captured and billed in the month-of-service. More than 25% were captured and billed in the 2 months following the month-of-service.

Line 03 shows the timing of the actual cash collections. Few practices collect significant income in the month-of-service because of the time it takes to capture the billing data and file claims. Also, insurance companies do not respond immediately to the claims. Their response strategy, unless contractually mandated, is influenced by cash flow agendas. Note that significant cash continues to be collected even after 23 months! The cash collected in the older cycles is attributed to claims involving litigation, or the billing office has negotiated installment plans with patients that have large balances. This technique allows conscientious patients to meet their obligations over a span that may cover 6 months to a year. In return, the billing office promises not to turn over the account to an agency.

Line 04 represents the contract adjustments required by the negotiated settlements with the payors. Payors will generally provide a document referred to as a remittance advice that accompanies the checks. They are similar in format to Table 1. The billing staff identifies both the cash payment and the contract adjustment from this document, and posts this information at the same time.

The last credit entry pertains to bad-debt write-offs (line 05). They do not generally occur until at least 5 months after the date of service. Most billing systems will first send a series of two to three detailed statements to patients that are responsible for the remaining account balance. If no response is received, they will be mailed a sequence of dunning letters warning that a failure to respond will result in the account being turned over to a collection agency. Most systems are programmed to automatically transfer delinquent self-pay accounts that have reached a certain age.

The volume of accounts necessitates this automatic process; it is often not cost-effective to pursue them, with the exception of those with very large balances. Keep in mind that significant resources have already been expended to this point in an attempt to effect payment. Billing vendors, or even internal systems, have to deal with cost/benefit trade-offs. These accounts are generally given to a collection agency that charges a higher fee to pursue the patient.

Lines 07 and 08 show the examination and account volume attributed to the gross charges. This information is used to compute the average cash income per examination and account (lines 13 and 14). The significance of this information will be made clear later in the article when we deal with the issue of system cost.

Lines 10 to 12 portray the benchmark collection ratios already discussed. By now you have observed that this database extends for 23 months. Most radiologists are not going to be interested in looking at the kind of information portrayed here. There are, however, benchmark data within this universe that can be organized in a manner that both educates the physician and offers a tool for rapid evaluation of billing consistency. We will next focus on how to measure relative time-sensitive performance.


We can take advantage of the powerful database features of today’s advanced systems by linking all transactions pertaining to a finite receivable universe (charges by month-of-service). It is possible to organize the benchmark ratios to measure consistency. A well-run system should capture and refine information, create patient accounts, file claims and bill self-pay balances, and complete the adjudication of accounts in a timing pattern that is easy to define. Tables 4, 5, and 6 illustrate how this can be done.

Table 3 offered the cumulative gross and net collection ratios at the end of each month following the month-of-service. The cumulative ratios represent a timing standard to visualize achieved results through a point in time. To measure consistencies, organize the cumulative ratios for all the monthly cycles in a manner similar to Table 4.

Each column represents the achieved cumulative ratio, starting with the month-of-service and continuing through month 18. The average of each column is at the bottom. Refer to the column with the heading Month 6. At the end of the sixth billing cycle of the month-of-service referred to as Month 1, the system achieved a cumulative gross collection ratio of 36.23%. In the sixth billing cycle of the month-of-service called Month 2, the system achieved a cumulative gross ratio of 40.88%.

Please recall, however, the discussions pertaining to the usefulness of gross collection ratios as a benchmarking tool. The data on Table 4 pertain to a single radiology practice. However, even within a single practice, there are going to be material differences in the third party population from one month to the next. We have to consider this when examining the differences in the achieved cumulative levels of each cycle. Also, because the information is organized by month-of-service, changes in a major payor’s fees can reduce the gross ratio. We would expect that the future patterns would show lower percentages than the previous cycles. For example, Medicare changes its rates every January. If Medicare fees decline significantly, it will not be possible for the future cumulative ratios to match the months preceding the change.

The percentages on each monthly line of Table 4 correspond to the line 10 calculation of Table 3. Table 5 shows the incremental gross ratios achieved in each cycle, measured against the total charges attributed to each month. Line 09 of Table 3 explains the computation. The averaging of the increment provides a useful statistic for forecasting. The sum of the averages is 51.71%. The normalization of this sum to 100% gives you the proportion of each month’s cash value that is collected in cycles 1 through >18. For example, the Month 1 value is .50/51.71=.98%. This value at the bottom of each column is referred to as average speed.

To illustrate the concept, Table 1 estimated the cash value of the $1,000 of gross charges to be $500.91. On average the practice would collect .93% of that in the Month 1 cycle, 25.88% in the Month 2 cycle, and so on.

We have found that these Table 5 ratios are relatively consistent from one practice to the next, and suggest to the readers that they can rely on them to be useful in estimating the cash flow of each month’s cash value.

Table 6 is the achieved cumulative net collection ratios through each cycle, and corresponds to line 11 of Table 3. The line 11 calculation is defined as:

Cumulative cash receipts/(Cumulative charges minus cumulative adjustments)

This is the most accurate way to measure processing consistency because you are removing the effects of fee schedule changes and payor reductions from the denominator. Also, this calculation minimizes material differences in the third party populations of each of the months. Therefore, you should expect the cumulative values in a column to be within a narrow standard deviation, especially in Month 6 forward. If the percentages in a column begin to decline, you have a processing problem.

There can be wide variations in the cumulative levels through the first 4 cycles. Then the patterns should begin to tighten around the mean, as evidenced by the use of standard deviation. In this specific case, the model highlighted difficulties with the earlier cycles where there was a lack of consistency. Since this was brought to the attention of the vendor, there has been an improvement in the achieved ratios for the most recent months.


If a practice meets with a billing vendor, and that vendor states that they have a standard processing rate of X%, find another one. All good companies understand the principles explained thus far. They will want to know your examination volume, but more important they will seek to estimate the number of new accounts created per year. Account management determines staffing, supply, and system needs. Companies that know their internal costs will have them measured on a unit basis. The most consistent measurement unit is account volume, especially in an automated environment.

The company will then use its knowledge of major payor reimbursement to estimate the cash value of an average account. Table 7 explains why. Let us return to the example of the four radiology practices, each billing 10 examinations. Line 01 portrays the average cash income per examination. Each practice covers a different hospital system with a different mix of inpatients and outpatients. The mix is important because outpatient accounts average between 1.15 and 1.40 examinations; inpatient accounts average 4.25. Practices A and C have a proportionately greater number of outpatient admissions than Practices B and D. Line 03 provides the cash value per account through the extension of income per examination and average examinations per account.

If there is a well-designed electronic interface between the receivable and hospital information systems, then it will take approximately the same resources to manage an inpatient- or outpatient-oriented account, even with material differences in the number of examinations posted to the accounts. The contractor knows that it needs to be paid a certain level per new account to cover direct expenses and generate a profit. The cash value of the account has no effect over this target level. The illustration uses a targeted fee per account of $6.50. Therefore, if a well-informed (and honest) vendor were processing the receivables of these four practices, it would charge four very different rates (line 05). These rates will produce different processing costs per examination, but this is irrelevant in a highly automated environment.


Our final points concern the two principal methods used to bill and collect for hospital-based radiology procedures. These are practice-owned and outsourcing. It will surprise many readers to learn that few practice-owned systems are as efficient as or less expensive than a good billing vendor. The general perception by radiologists has been that billing vendors have to be more expensive because they have a profit margin built into their rates. Radiologists also feel that they exercise more control over their own systems and have more dedicated staff.

Our examinations of both types of methods have not convinced us that internal systems offer any tangible advantages. They actually tend to be more expensive because they do not have the economies of scale and technological sophistication of the vendors, with a tendency toward higher staffing costs.

Internal systems are now becoming riskier because of increased compliance demands. The combination of higher processing fees and constant pressure by payors for price concessions makes it difficult to rationalize future commitments for system upgrades that cost, even for a small practice, many tens of thousands of dollars. Radiologists with in-house billing systems may want to consider shifting the risk of future uncertainties to a specialized vendor, especially when it is likely that costs may actually go down and collection efficiency will either be the same or better.

James A. Kieffer is president of Proforma Financial Group Inc, a health care consulting firm in Concord, MA.