What is the difference between an LCD and an NCD?

An NCD (National Coverage Determination) applies to all Medicare contractors nationwide — it sets the floor for coverage. An LCD (Local Coverage Determination) is issued by a specific Medicare Administrative Contractor and applies only to providers in that jurisdiction. LCDs can be more restrictive than NCDs but cannot override them. When the tool cross-references coverage policies, it checks both — and flags when a service is covered under the NCD but the LCD imposes additional documentation requirements.

Can it check coverage for both Medicare and commercial payers?

The tool cross-references LCDs and NCDs, which are Medicare-specific. For commercial payers, the AI analyzes the clinical documentation against the denial or coverage information in the uploaded documents. Commercial payer policies vary widely and are generally not published in a standardized format — so the tool is most effective for Medicare prior auths, and serves as a documentation quality check for commercial submissions.

How long does the AI analysis take?

Typically 30-60 seconds for a standard PA request with supporting records. The processing time depends on the volume of uploaded documentation and the complexity of the coverage cross-reference. The report appears on screen as structured markdown as soon as the analysis is complete.

What happens if the AI identifies a documentation gap?

The report flags the gap explicitly — what's missing, why it matters, and what specific documentation would address it. For example, if the LCD requires a specific lab value within the last 30 days and the uploaded records don't include it, the report will say exactly that. You can then gather the missing document and resubmit, or upload additional records and re-run the analysis.

Is patient history required or optional?

Optional, but strongly recommended. The PA request alone gives the AI the procedure codes and the basic clinical rationale. Adding the patient history — imaging reports, lab results, procedure notes — allows the AI to cross-reference the clinical evidence against the coverage criteria and identify gaps you might not have noticed.

Is this a substitute for a utilization review specialist?

No. The tool is a productivity aid that accelerates the coverage research and documentation gap analysis. It does not replace clinical judgment, payer negotiation, or the expertise of a certified utilization review professional. Use it to make your UR team faster and more thorough, not to replace them.

Guide · 8 min read

Prior Authorization:
A Step-by-Step Guide

Reduce auth delays with smarter coverage checks and stronger documentation. Real advice from someone who has reviewed more PA requests than most payers process in a year.

Try the Prior Authorization Tool

The Short Version

The Prior Authorization tool takes a PA request and the patient's clinical history and runs it through a coverage policy check, code validation, and documentation gap analysis. Upload the auth form or clinical documentation, add supporting records, and the AI cross-references the requested procedure against applicable LCDs and NCDs.

The output is a structured PA review report that tells you whether the service is likely covered, whether the submitted codes support medical necessity, and what documentation is missing — before you submit to the payer. It is the second set of eyes every auth packet deserves but few practices have the time to do.

90%

Providers Report Auth Burden

Input Sources

Export Formats

100%

Free. No Account.

Why Prior Authorization Is Broken

I once watched a three-vessel CABG get delayed by three weeks because the prior auth packet included the procedure codes and nothing else. No clinical narrative. No stress test results. No documentation of medical therapy failure. The payer didn't deny it — they just sat on it. When the surgeon's office called to check the status, the response was: 'We need more clinical information.' Three weeks. For a coronary bypass.

That story is distressingly common. The American Medical Association reports that 90% of physicians describe the prior authorization burden as high or extremely high. The average practice spends two business days per week per physician on PA-related tasks. And one in seven PA requests is initially denied — often for reasons that could have been addressed before submission.

The problem is not that prior authorization is unnecessary. When used appropriately, it ensures that services are medically necessary and appropriately coded before they are performed. The problem is that the process is fragmented across hundreds of payers, each with their own forms, criteria, and timelines — and most practices don't have the staff to navigate each one individually.

A strong prior auth submission is the single best way to avoid a denial before it happens. It means understanding the coverage criteria before you submit, validating that your codes and documentation meet those criteria, and identifying gaps while they are still easy to fix.

How the Prior Authorization Tool Works

Upload the Prior Auth Request

Upload the prior authorization request form, CMS-1500, UB-04, or any clinical documentation that describes the proposed service. The AI reads the request and identifies the procedure codes, diagnosis codes, and the clinical rationale.

AI Checks LCD & NCD Coverage Policies

The agent cross-references the requested procedure against applicable Local Coverage Determinations and National Coverage Determinations. It identifies whether the service is covered under the patient's benefit category and flags any coverage criteria that need to be met.

AI Validates Codes & Flags Documentation Gaps

The AI checks that the submitted diagnosis codes support medical necessity for the requested procedure, validates code pairs for correct coding initiative compliance, and flags missing documentation elements that would trigger a denial or delay.

Review the PA Review Report

The generated report shows the coverage determination, code validation results, documentation gap analysis, and recommendations. Download as Markdown, HTML, or PDF and attach to your auth submission for a stronger packet.

Common Prior Auth Scenarios

MRI or CT Imaging Prior Auth

An MRI of the lumbar spine is ordered for a patient with chronic lower back pain and radicular symptoms. The payer requires prior authorization with specific clinical criteria — failure of conservative therapy, specific physical exam findings, and a documented duration of symptoms. The AI reads the clinical documentation, confirms that the patient has tried 6 weeks of physical therapy and NSAIDs (meeting the conservative therapy requirement), validates the ICD-10 code for lumbar radiculopathy (M54.16) against the CPT code for lumbar MRI (72148), and flags that the physical exam findings are documented only in a narrative format rather than a structured template.

Surgical Procedure Authorization

A cardiologist requests prior authorization for a cardiac catheterization (93458) based on abnormal stress test results. The payer's LCD requires specific documentation: the stress test report, the EF percentage, and the Canadian Cardiovascular Society angina class. The AI reviews the uploaded records, confirms the CPT-LCD match, identifies that the CCS angina class is documented but the exact EF percentage is not explicitly stated in the chart notes — and flags it as a documentation gap that could delay the auth.

Specialty Medication Prior Auth

A rheumatologist requests prior authorization for a biologic agent (like adalimumab) for a patient with moderate-to-severe rheumatoid arthritis. Specialty medication PAs have notoriously strict criteria — failure of two DMARDs, specific lab values, and disease activity scores. The AI cross-references the requested medication against the applicable NCD and commercial payer criteria, validates that the patient has documented failure on methotrexate and sulfasalazine, flags that the most recent DAS28 score is missing, and recommends including it before submission.

Tips from the Trenches

Three things I wish every provider knew before submitting a prior auth.

Timing Is Everything — Too Early Is Almost as Bad as Too Late

Most payers have a window for prior authorization: typically 30 to 90 days before the planned procedure. Submit too early and the auth expires before the procedure date, requiring a new submission from scratch. Submit too late and you risk a denial or a delay. I recommend submitting 2-3 weeks before the scheduled service — enough time for the review and any appeals, but not so early that the auth window closes. Mark the expiration date on your calendar the day the auth comes back.

The Clinical Evidence Bundle Is Your Strongest Argument

A prior auth request is not a form — it's a medical necessity argument. The strongest auth packets include three things: (1) the clinical narrative explaining why this service is needed for this patient at this time, (2) the supporting objective evidence (lab results, imaging reports, pathology), and (3) the documentation of failed conservative therapy if applicable. I've seen auths that were identical in every way except one included a one-paragraph clinical rationale and the other didn't. The one with the rationale was approved in 24 hours. The other was denied for 'insufficient clinical documentation.'

Concurrent Review Can Save You from a Stop-Loss Situation

For high-cost services or extended inpatient stays, concurrent review — submitting clinical updates during the course of treatment — can prevent a retrospective denial that leaves you holding the bag. Many providers treat prior auth as a one-and-done event, but payers increasingly require ongoing documentation for services that span multiple days or involve multiple procedures. I tell every facility I consult with: if the expected length of stay exceeds three days, have your utilization review team check the auth status on day two.

Frequently Asked Questions

Questions I hear from practices navigating the prior auth process.

Ready to Speed Up Your Prior Auths?

No sign-up, no credit card, no upsells. Upload a PA request and see what the AI finds in seconds.