3107 — Do Prevention Quality Indicators Underestimate Preventable Hospitalizations?
Tseng C, VA New Jersey Health Care System; Soroka O, VA New Jersey Health Care System; Pogach L, VA New Jersey Research Office;
Agency for Health Research and Quality (AHRQ) Prevention Quality Indicators (PQIs) are measures of quality of care for "ambulatory care sensitive conditions" from hospital discharge data. However, there are additional hospitalizations that can be prevented by appropriate ambulatory care and may be considered to be added to the list of PQIs. Our objective was to evaluate rates of five AHRQ PQIs (metabolic decompensation (MD): short term diabetes complications, dehydration, and uncontrolled diabetes), lower extremity (LE) amputation (LEA), and urinary tract infection (UTI)), and three proposed preventable hospitalizations (PHs) (lower extremity ulcers/infections (LEU), hypoglycemia (HYPO), and Acute Kidney Injury (AKI)).
Using Veterans Health Administration (VHA) diabetes patients in 2006 combining VHA and Medicare records, we calculated PH rates (per 1,000 persons) and determined age-and-sex adjusted rates by race. We used principal ICD-9-CM codes for the proposed PHs: 1) LEU (Harrington C. 2000; excluding co-occurring LEAs)); 2) AKI: code 548.9; and 3) HYPO (Ginde 2008), and AHRQ technical specifications for current PQIs.
There were 975,667 individuals; 57 % were 65 years or older, 97% men, 76% Whites, and 13% each for Blacks and other racial groups (others). Proposed PHs (17,069 admissions) comprised 47% of all PHs (35,943 admissions). AKI was the most common PH (8,507 admission; 8.7 admissions per 1,000 persons), followed by MD (7,500; 7.7), UTI (6,130; 6.3), LEA (5,244; 5.4), HYPO (4,979 ; 5.1), and LEU (3,583; 3.7). LE PHs increased by 41% by including LEU, and MD by 40% by including HYPO. Blacks were more likely to incur all/AHRQ/proposed PHs: 55/28/27 (per 1,000 persons) vs. 35/18/16 (others) vs. 39/20/19 (Whites). LEU added to the number/rate of LE PHs more in Whites (42%) and others (42%) than Blacks (33%); HYPO added to MD more in others (47%) and Blacks (46%) than Whites (38%). The comprising percentages of three proposed PHs in all studied PHs were similar in Whites (47%), others (49%), and Blacks (49%).
Current PQIs missed about half of diabetes relevant PHs. The magnitude of underestimation differed for individual PHs by race.
VHA should consider including additional PQIs to its existing PQI measurement system