21.2 PSI 90 Components and PSIs beyond the Composite

21.2 PSI 90 Components and PSIs beyond the Composite

Quick Answer: The CMS Patient Safety and Adverse Events Composite (CMS PSI 90) used in the Hospital-Acquired Condition Reduction Program (HACRP) currently includes ten Agency for Healthcare Research and Quality (AHRQ) Patient Safety Indicators (PSIs): PSI 03 pressure ulcer; PSI 06 iatrogenic pneumothorax; PSI 08 in-hospital fall-associated fracture; PSI 09 postoperative hemorrhage or hematoma; PSI 10 postoperative acute kidney injury (AKI) requiring dialysis; PSI 11 postoperative respiratory failure; PSI 12 perioperative pulmonary embolism (PE) or deep vein thrombosis (DVT); PSI 13 postoperative sepsis; PSI 14 postoperative wound dehiscence; and PSI 15 abdominopelvic accidental puncture or laceration. A diagnosis with present on admission (POA) = Y is treated as not hospital-acquired and is generally excluded from the numerator. Other PSIs exist outside this composite; do not invent an outdated full list or restore PSI 07 as if it still sat inside PSI 90.

This independent OpenExamPrep section helps inpatient clinical documentation integrity specialists study how CMS PSI 90 reads the coded Inpatient Prospective Payment System (IPPS) claim. It is not a CMS, AHRQ, or Association of Clinical Documentation Integrity Specialists (ACDIS) reprint. Section 21.1 placed PSI 90 inside HACRP. This section names the components, the POA gate, and the boundary between the composite and the rest of the PSI family.

What PSI 90 is—and what it is not

AHRQ developed PSIs as administrative-data screens for potentially preventable in-hospital complications. CMS calculates CMS PSI 90 from Medicare fee-for-service claims using PSI specifications (CMS’s implementation of the AHRQ software for the HACRP / public-reporting versions in use that year). The composite is a weighted combination of component observed-to-expected ratios. AHRQ publishes harm- and volume-based component weights in its PSI composite documentation; those weights change by software version. Do not memorize a weight table as if it were a CMS unpublished cut score, and do not hand-calculate a hospital’s composite on the exam.

PSI 90 is not:

• The 14-category hospital-acquired condition present-on-admission (HAC-POA) list (claim grouping).
• National Healthcare Safety Network (NHSN) infection surveillance (chart/laboratory/device protocols).
• A license to drop a valid diagnosis from the claim “so PSI does not see it.”
• Chart-abstracted severity of illness (SOI) / risk of mortality (ROM).

It is a claims composite. If the diagnosis is not coded, or is coded without the procedure, exclusion, or POA the specification requires, the software never sees the event. If the diagnosis is coded and matches the numerator with POA = N, the software can see it even when nobody in quality review has opened the chart yet.

The current ten: CMS PSI 90 components

CMS’s live HACRP description and current AHRQ PSI 90 composite documentation (v2025) name the same ten indicators. Learn numbers, names, and the documentation question each one asks. The ACDIS handbook shorthand “PSIs 03, 06, and 08–15” is the same set; it is not “03 through 15 including 04, 05, and 07.”

PSI 08 name change is a trap. Older materials said postoperative hip fracture. Current AHRQ/CMS language is in-hospital fall-associated fracture. The event is no longer limited to a postoperative hip. A medical patient who falls on day 4 and fractures a femur can belong to PSI 08. A patient admitted with a hip fracture cannot.

PSI 10 name change is a trap. If a stem still says “physiologic and metabolic derangement,” translate it to postoperative AKI requiring dialysis for today’s composite, or recognize the stem as stale.

POA = Y means not hospital-acquired for PSI

PSI software is hunting in-hospital events. The POA indicator is the usual gate:

• POA = Y: the condition was present at the inpatient admission order. For PSI 90, that diagnosis is generally excluded from the numerator as not hospital-acquired. Conditions that arise in the emergency department, observation, or outpatient surgery before the inpatient order remain Y—the same clock Domain VI taught.
• POA = N: not present at the inpatient order. If the diagnosis/procedure pair matches a component, the stay can enter that numerator.

Do not copy the HAC-POA payment table onto PSI software. Under the Deficit Reduction Act provision, N and U generally block complication or comorbidity (CC) / major CC (MCC) grouping from a selected HAC, and W still pays like Y. That table answers MS-DRG money on this claim. PSI 90 answers whether a quality composite counted a hospital-acquired safety event. If an item asks only about PSI exclusion, the high-yield answer is POA = Y excludes as not hospital-acquired. If it asks about HAC-POA payment, use N/U versus Y/W. Mixing those two sentences is how people miss both items.

U (documentation insufficient) is still a documentation failure. It is a reason to query for timing, not a reason to invent how every vendor treats U inside every PSI version. W (clinically undetermined) is the provider’s inability to determine timing. Neither letter is a quality workaround.

Worked PSI encounters

Community Stage IV ulcer. Sacral full-thickness ulcer photographed in the emergency department before the inpatient order; attending later writes Stage IV. POA = Y. This is not a PSI 03 hospital-acquired numerator. It may still be an MCC. It may still be clinically severe. PSI 03 does not “take community ulcers.”

Hospital-acquired Stage IV. Admission skin exam intact; day 12 heel ulcer documented Stage IV. POA = N. This can enter PSI 03 if the code is in the spec. Separately, HAC-POA may ignore it for CC/MCC grouping. Two ledgers, one ulcer.

Admission hip fracture versus inpatient fall. Patient admitted for femoral-neck fracture after a fall at home: principal fracture, POA = Y, not PSI 08. Patient admitted for pneumonia, falls on day 3, new humeral fracture documented: POA = N, PSI 08 candidate under the fall-associated fracture definition.

Iatrogenic pneumothorax. New pneumothorax after subclavian central-line attempt, POA = N: PSI 06 candidate, and it may also be HAC-POA category 14 if the codes pair. New pneumothorax after lung biopsy: still PSI 06 if coded as iatrogenic and not POA; not automatically HAC-POA category 14.

Postoperative dialysis. Elective colectomy, creatinine 0.9 on admission, anuric on day 2, new hemodialysis for AKI. PSI 10 candidate. Patient already on chronic dialysis for end-stage kidney disease, creatinine unchanged: that is not new postoperative dialysis for AKI.

Sepsis timing. Sepsis with organ dysfunction present in the emergency department, then emergency laparotomy: postoperative PSI 13 is the wrong reflex. Sepsis first documented on postoperative day 4 after an elective case, POA = N: now you are in PSI 13 territory if codes and exclusions match—still query for clinical validation, not for “please prevent a PSI.”

PSIs beyond the composite—without a fake master list

AHRQ’s PSI module is larger than the ten indicators inside PSI 90. CMS has also publicly reported selected PSIs that are not in the HACRP composite—most notably PSI 04, death rate among surgical inpatients with serious treatable complications (often taught as failure to rescue). Other module indicators have included death in low-mortality DRGs, retained surgical item counts, and obstetric trauma indicators. Those tools matter to quality departments. They are not this guide’s invitation to freeze a historical inventory and call it 2026 HACRP.

Do not restore PSI 07 (central-line-associated bloodstream infection on claims) to the composite. CMS removed that claims infection indicator from PSI 90; HACRP now uses NHSN CLABSI instead of asking PSI 90 to double-count line infections. An item that lists PSI 07 among “today’s PSI 90 components” is using an outdated composite.

Do not equate PSI 05 (retained surgical item) with HACRP PSI 90 just because retained object is also a HAC-POA category. Different engines.

When a stem asks “which PSIs are in the CMS composite used for HACRP,” answer 03, 06, and 08–15 with the current names. When it asks whether other PSIs exist, the correct idea is yes, outside the composite and in some public reports, not a recitation of every retired label.

What documentation work changes PSI 90

• Code the clinically valid diagnosis when Uniform Hospital Discharge Data Set (UHDDS) additional-diagnosis rules are met. Omitting a Stage IV ulcer to “protect PSI 03” is a compliance problem.
• Assign POA against the inpatient order, including pre-admission hospital time.
• Query staging, acuity, postoperative relationship, and accidental versus expected injury when the record is unclear—without reimbursement or quality-outcome language (2026 query brief).
• Know spec limits: PSI 12 ≠ HAC-POA hip/knee DVT only; PSI 06 ≠ HAC-POA venous-catheter pneumothorax only; PSI 08 ≠ postoperative hip only.
• Leave NHSN case-finding to infection prevention’s protocol. A PSI 13 code path is not an automatic NHSN SSI, and the reverse is also true (next section).

Traps this section is built to catch

• Treating PSI 90 as the 14 HAC categories or as NHSN.
• Using POA = Y as if it still counted as hospital-acquired for PSI.
• Importing HAC-POA N/U payment blocking as the PSI software rule.
• Calling PSI 08 postoperative hip fracture only.
• Calling PSI 10 any postoperative electrolyte disorder.
• Listing PSI 04, 05, and 07 inside today’s HACRP composite.
• Inventing a complete outdated PSI catalog and presenting it as current CMS policy.
• Querying “please document that this was POA so we do not get a PSI.”

The composite sentence is short: CMS PSI 90 for HACRP is 03, 06, and 08–15. POA = Y excludes the diagnosis as not hospital-acquired. Other PSIs exist outside that composite; do not fake the rest of the list.