Patients with diabetes are frequently hospitalized for potentially preventable hospitalizations (PHs). These are operationalized by the Agency for Health Research and Quality (AHRQ) as Prevention Quality Indicators (PQIs). However, other PHs relevant for diabetes patients (hypoglycemia (HYPO), lower extremity ulcers/inflammation/infections (LEU), and acute kidney injury (AKI)) have not been substantially evaluated.
The study aimed to evaluate among patients diabetes (and CKD):
(1) Whether current PQIs underestimate disease burden.
(2a) Rates/trends of current and proposed PQIs.
(2b) Rates and trends of serious hypoglycemia (hospitalizations and emergency department (ED) visits).
(2c) Rates and trends of initial and repeat lower extremity amputations (ILEA and RLEA).
(3) Two different algorithms for identifying AKI.
(4) Utilization of subspecialty and primary care.
We performed secondary analysis of data files from Veterans Health Administration and Medicare on Veterans with diabetes and CKD.
Objective(1). Utilized serial cross-sectional cohorts of patients with diabetes and CKD (stages 2-5) in fiscal years(FYs) 2000-2004. AHRQ PQIs studied were metabolic decompensation (short-term diabetes complications, dehydration, and uncontrolled diabetes without complications), cardiovascular conditions (hypertension, congestive heart failure, and angina without procedure), lower extremity amputation (LEA), and infections (bacterial pneumonia and urinary tract infection (UTI)); and three proposed PQIs (HYPO, LEU, and AK)I.
Objective(2a). Among diabetes patients in calendar year (CY) 2012 aged >=65 years, we assessed seven existing PQIs (four from AHRQ Diabetes Composite: uncontrolled diabetes, short-term complications, long-term complications, LEA; three from Acute Composite: UTI, bacterial pneumonia, dehydration) and the three proposed PQIs to create an expanded Diabetes Composite.
Objective(2b). Utilized serial cross-sectional cohorts of diabetes patients in CYs 2000-2007. Serious hypoglycemia was assessed by subpopulation.
Objective(2c). Utilized a serial cross-sectional design. We assessed LEA rates in CYs 2000-2012 by type.
Objective(3). We used the principal diagnosis code 584.9 (ADMIN) as well as combination of ratio and difference from two serum creatinine values (LAB) to determine AKI in CYs 2006, 2009, and 2012 among diabetes patients.
Objective(4). Among diabetes patients in CKD stage 3-5 in CYs 2000-2012, we assessed utilization of subspecialty in nephrology, endocrinology, and cardiology and primary care.
Objective(1). The study cohort increased from 155,081 in FY2000 to 255,638 in FY2004 (65% increase), with 32% having stage 3-5 CKD. There were consistent downward trends of various degrees (19%-62%).
The rate of any PH was 103.1 in FY2000 and 73.7 per 1,000 patients in FY2004, a 28.5% decrease. Consistently over the years, congestive heart failure and AKI were the top two PHs; LEU and hypertension were the lowest two.
Objective(2a). In 2012 there were 792,530 diabetes patients aged>=65 years; patients with complex comorbid conditions (advanced diabetes complications, cancers, major neurological conditions, major depression, substance/alcohol abuse) comprised 29%. The numbers/rates (per 1,000 patients) of total, existing and proposed PQIs were 48,260/60.9, 32,217/40.7 and 18,443/23.3. Existing PQIs did not capture one-third of diabetes relevant PHs, similarly across studied racial/ethnic groups. The Acute Composite increased by 30% by adding AKI. An expanded Diabetes Composite had > 4.5 fold increase in the rate than the AHRQ Diabetes Composite (60.9 vs. 13.0). Blacks and patients with complex comorbid conditions had higher rates than other subpopulations.
Objective(2b). In years 2000 and 2007, 479,081 and 1,010,095 patients had diabetes; 66% and 52% received insulin and/or sulfonylurea (I/SU). There were consistent decreases in rates for serious hyperglycemia (overall relative reduction in diabetes and I/SU populations: 44% [577-321/100,000 patients] and 36% [673-432] with hospitalizations; 31% [784-545] and 22% [897-701] with ED visits without hospitalization) and hospitalization for hypoglycemia (14% [510-440] and 2% [633-620]), while rates of ED visits for hypoglycemia continued to increase (22% [1287-1573] and 41% [1569-2212]).
Objective(2c). There were 547,382/1,320,758 diabetes patients in 2000/2012 without prior amputations, incurring 3,246/3,250 amputations. ILEA rates decreased by 59% (5.93 in 2000 to 2.44 in 2012); major and minor ILEA rates decreased by 67% (2.10 to 0.69) and 54% (3.82 to 1.77), respectively. The ratio of major-to-minor ILEA rates reduced from 0.55 in 2000 to 0.39 in 2012. There were 7,928/9,742 diabetes patients in 2000/2012 with amputation procedures in prior two years, incurring 1,011/1,135 amputations. RLEA rates decreased in a linear fashion (127.5 in 2000 to 118.7 in 2012; 7% reduction). Major RLEAs decreased by 23% (59.79 to 45.91). The major-over-minor RLEA ratios decreased across years (0.88 to 0.63). Blacks and Whites both experienced linear downward trends in ILEA and RLEA.
Objective(3). AKI increased from 2006 to 2012 (ADMIN: 3,370 to 5,504; LAB: 26,904 to 29,979). Kappa statistics for dis/agreement of AKI ranged from 0.12-0.15. Of the AKI identified by either approach (27,008-30,157); 93-95% were identified from LAB only, and <1% were from ADMIN only. The AKI rate slightly decreased by 2% from 2006 to 2012 (27.5 to 26.19/1,000 patients).
Objective(4). The percentage of diabetes patients with stage 3-5 CKD who made visits to selected specialists (nephrologists, endocrinologists, cardiologists) increased from 38% in 2000 to 49% in 2012. Cardiologists were visited most often (26%; 34%), followed by nephrologists (15%; 23%) and endocrinologists (6%; 8%). However, in our measure of consistency of care, patients who made visits in 3 or 4 quarters in a year remained low (8% in cardiology, 7% in nephrology, and 3% in endocrinology in 2012). Among those in CKD stage 4, only 26% visited nephrologists in three or four quarters in 2012, compared to 21% in 2000.
Our results demonstrate the feasibility to improve adverse outcome surveillance in dually enrolled Veterans using operational data.
Objsectives(1 & 2a). VA can consider adding additional diabetes specific and diabetes relevant PQIs to evaluate the burden of PH.
Objective(2b). Surveillance of serious hypoglycemia and hyperglycemia should report both ED and H.
Objective(2c). There should be separate reporting of ILEAs and RLEAs in order to monitor the progress in decreasing amputation rates.
Objective(3). It was feasible to use LAB data determining trending AKI using electronic health records
Objective(4). A focus upon Stage 4 CKD patients is warranted.
- Tseng C, Soroka O, Pogach LM. Serious Hypoglycemia and Hyperglycemia Events in the Veterans Health Administration (VHA) in 2000-2007. [Abstract]. Diabetes. 2014 Sep 1; 63(Suppl 1):A46.
- Tseng C, Soroka O, Lafrance JP, Pogach L. Prevalence of Acute Kidney Injury (AKI) Is Underestimated Using Administrative Data. Presented at: American Diabetes Association Annual Scientific Session; 2015 Jun 1; Boston, MA.
- Tseng C, Soroka O, Pogach L. Do Prevention Quality Indicators (PQIs) Underestimate Preventable Hospitalizations (PHs)? Presented at: American Diabetes Association Annual Scientific Session; 2015 Jun 1; Boston, MA.
- Tseng C, Soroka O, Pogach LM. Serious Hypoglycemia and Hyperglycemia Events in the Veterans Health Administration (VHA) in 2000-2007. Paper presented at: American Diabetes Association Annual Scientific Session; 2014 Jun 15; San Francisco, CA.
- Tseng C, Soroka O, Lafrance JP, Maney M, Pogach L. Preventable Hospitalizations in Patients with Diabetes and Chronic Kidney Disease. Paper presented at: AcademyHealth Annual Research Meeting; 2013 Jun 24; Baltimore, MD.