A receivable system audit is a statistically significant testing of critical components to ascertain billing efficiency and integrity that is generally conducted by a party with no legal or financial relationship to the organization that performs the billing function. Most auditing protocols apply to all medical receivable systems. Billing the interpretation of radiology procedures performed in a hospital has certain unique requirements. This stems from lack of control over ordering examinations, lack of direct access to patients, and reliance on the hospital for clinical and demographic information.
An important issue for a practice that has concerns about the efficiency of its receivable system is the cost of audit testing in light of what is already paid for the billing function. Radiology systems process many more new accounts per month than every other specialty and likely pay more in gross dollars for this function than even the largest office-based specialists. A “certifiable” audit is a lengthy, manual process. The composite senior and junior staff time per account is extensive.
The first half of this article will focus on the considerable resources required to conduct a conventional, manual audit. The second part will describe an alternative mthod that harnesses the powerful database capabilities of modern receivable systems.


The following data are critical to conducting a thorough manual audit.

  1. Secure, at the hospital, all evidence of performance of clinical service in the form of written finding(s) that include reference to the ordering physician, date/time the image was obtained, date/time of radiologist dictation, and radiologist name. Confirm that the hospital has posted the technical charge to its general ledger (the hospital billed it because the hospital imaged the patient).
  2. Expedite assignment of appropriate diagnosis (ICD-9) and procedure (CPT-4) code(s) based on clinical information evident on written findings.
  3. Secure, from the hospital, demographic information for the patient that provides evidence of coverage for health care services.
  4. Request a hard copy of the history file(s) electronically resident in the receivable system. It will be either a single (very crowded) data sheet or multiple forms. The critical data fields concern patient identity, date of service (DOS), responsible party(s), ICD-9/CPT-4 code(s), date the transaction was posted (DOP) to the system, fees consistent with procedure code(s), dates(s) and type(s) of claims/statements, date(s) and type(s) of cash and noncash credits.
  5. Validation of actual billing initiatives is one of the more difficult areas of auditing because computerized systems will show dates of bill/claim submission but there may not be physical evidence on-site. Seek out copies of production logs that list patients included in bill/statement batches. Only as a last resort seek contact with patients/insurers to confirm their receipt of statement(s)/claim(s).
  6. Secure hard copy(s) of all payor remittance advice(s).
  7. Secure hard-copy documentation of actual payment medium and deposit backup that foot to the remittance advice(s).


The best vehicle to document findings from a comprehensive manual audit is a database program; spreadsheets are also suitable. It is not the purpose of this article to explain how databases work. The only descriptions of note are the distinction between a table, record, and field. A table consists of records (rows) and fields (columns). Databases often have multiple tables, where each has specialized field information. The data in a field is identical in type and format. It is essential that each table contain common fields to facilitate merging of information.

Table. Estimates of the cost of a comprehensive audit based on examination volume reviewed.

There will be 40 to 50 fields that include all transaction codes and dates from the system records and fields populated with information that differs from system data. Two important fields are payor codes and payor fees by CPT-4. This enables the auditor to isolate records specific to that payor to inform the client how this population was managed.


Once the data has been organized into a database program, the user can organize information in a manner that isolates important trends, thereby generating a variety of reports. Below are tests that can be run on audit data:

  1. Was a record for this hospital patient resident on the billing system?
  2. Do the system CPT-4/ICD-9 codes match up with those independently assigned? If not, how do the charges differ and what might this mean in cash revenue?
  3. What are the average differences in hurdle dates: DOS versus DOP, DOP versus date of claim (DOC), DOC versus date of receipt (DOR), DOR versus DOP?
  4. Are there differences in the CPT-4 code on the claim versus the remittance advice? Does it produce income shortfall?
  5. Are there differences in payment/CPT-4 versus known payor fee schedule?

Another by-product of database compilation is payor-specific collection ratios (gross and net) for an audited population. This database is linking the cash and noncash credits to the original charge. The population, if random enough, is a microcosm of the entire receivable system. Reconciliation benchmarks can be measured against the standard system reports.

The table above offers estimates of the cost of a comprehensive audit.

The headings explain the step-down, beginning with monthly examination volume simulating a range of practice sizes, or a specific payor population within a system. A client may wish only to test a major payor where the fee schedule relationship is well documented. Using average examinations per account (inpatient and outpatient) of 2.15 tells us the monthly account volume. A 10% random sample size is large enough to draw conclusions about the entire population. The aggregate time to capture, document, and report on all information will be 45 minutes per account. These time commitments translate to fees based on a range of average hourly rates for junior and senior auditors, where the majority of hours will be data collection. The senior auditors will interpret the data and report the findings.


The better receivable systems have powerful database capability that can be used to build very large files of field information similar to the log created from the manual construction of the 10% account sample. The size of this sampling can include an entire month’s caseload. The size of the practice is irrelevant.

There are compromises because of reliance on the quality and integrity of the system data. Random testing can validate the field information in this file in a manner similar to the manual audit:

  1. Choose 100 records from the file, regardless of the size of the practice. Each record contains all relevant fields of transactional and date information.
  2. Secure the source information from the hospital for these patient records to verify relevant clinical and demographic field data.
  3. Secure hard copies of the remittance advices and all deposit information to confirm levels of payment(s), credits(s), and deposit(s) dates.

This back testing, even though a small sample, should confirm that the file data is representative of how the system handled this large population of records. If this test reveals major differences between file data and hard-copy documentation that alone is a significant finding. However, it also undermines this approach as a credible representation of system efficiency/integrity.

It is possible to produce exactly the same models as the smaller manual population, including the payor-specific collection performance. As with the manual audit, the two most critical information fields will be accurate payor codes and the contracted fees of the important payor(s). This enables the auditor to screen examinations performed on patients covered by the payor(s) in the month. A variance report will isolate records where payment is less than payor fee.

There are practical impediments to screening actual payments against payor fee schedules. It is often not possible for the billing system to segregate accounts by payor contract. Major payors generally have three product lines in the same market (HMO, PPO, and POS). Each has a different fee schedule. The billing system will not know in advance which product the patient has; all accounts will have the same payor code. The screening would require a “process of elimination” methodology. The first pass will produce match-up hits against one product fee; those records are then segregated from the rest of the population. The remaining records are then screened against the next product fee schedule with a second set of match-up hits, leaving the last group to be screened against the final product. The records remaining from this process are those that “might” be wrong for some reason. These are the examinations requiring acquisition of hard-copy documentation.

The final point concerning this technique is that, for very large populations of examinations, it is half as expensive as the manual audit while testing entire monthly populations. This technique will not isolate procedure coding errors or determine if there is a procedure capture problem, two essential steps of an audit. Those components of testing should be done with the 10% population target.


The traditional manual audit of a statistically significant population of radiology examinations is expensive and only appropriate if there is suspicion of severe system deficiency. The database features of most modern receivable systems allow an auditor to secure entire monthly populations of records where the charges and related cash and noncash credits can be linked together for payor-specific analysis. This can be done at about one half the cost and may actually provide more statistically significant information.

James A. Kieffer, MBA, is president, ProForma Financial Group, a consulting service in radiology reimbursement based in Nashua, NH, (603) 598-2944