Proposed Changes to Existing Measure for HEDIS 2015: Plan All-Cause Readmissions (PCR)

Transcription

1 Draft Document for HEDIS 2015 Public Comment Obsolete After March 19, Proposed Changes to Existing Measure for HEDIS 2015: Plan All-Cause Readmissions (PCR) NCQA seeks comments on proposed modifications to the Plan All-Cause Readmissions measure. This measure assesses the number of hospital discharges that were followed by a readmission for any diagnosis within 30 days, and the predicted probability of an acute readmission. We propose the following changes: 1. Eliminate the exclusion that removes readmissions from the denominator, so readmissions can serve as a potential index admission. 2. Add an exclusion for planned readmissions. The proposed changes will improve the measure s validity and align it with the CMS Hospital Wide Readmission measure, which assesses hospital-level performance at reducing readmissions for Medicare fee-for-service beneficiaries. Eliminating the exclusion for readmissions from the denominator ensures that the measure captures every readmission, even if there are multiple readmissions. The exclusion for planned readmissions ensures that the measure captures only unplanned readmissions. We tested these proposed changes in a research database of two years of Medicare Advantage health plan claims data with more than 1 million hospital admissions, provided by Inovalon, and two years of commercial claims data with more than 1 million hospital admissions, provided by OptumInsight. Data from one year are presented below for brevity, although results were similar in both years of data. Excluding planned readmissions from the numerator reduced the rate of readmission, while allowing readmissions to count as index admissions increased the rate of readmission. Allowing readmissions to count as index admissions had a greater effect, leading to an overall increase in the observed rate of readmission (an average 12 percent increase for commercial beneficiaries; a 5 percent increase for Medicare beneficiaries 65 [Table 1]). We estimated a new risk adjustment model for expected readmissions using the revised approach to defining hospital stays and readmissions. On average, plans show almost no change in their observed to expected ratio (O/E) with the proposed approach. A simulation of results in a sample of plans showed that some plans may see their O/E increase slightly, while others will see their O/E rate decrease slightly. Results were similar for 2011 and Table 1: 2011 Change in Observed Readmission Rate and O/E Ratio With Proposed Approach Current PCR Approach: Readmission not as an index admission and planned admissions as readmissions Revised Approach: Readmission as an index admission and planned admissions not as readmissions Observed Readmission Rate Commercial Avg. O/E (n=72 plans) Observed Readmission Rate Medicare 65 Avg. O/E (n=47 plans) 10.60% % % % 0.96 Avg O/E: Average observed rate of readmission/expected rate of readmission. Supporting documents for the proposed measure include the draft measure specification, evidence work-up, and performance data. NCQA acknowledges the contributions of the Geriatric Measurement Advisory Panel.

2 Draft Document for HEDIS 2015 Public Comment Obsolete After March 19, Plan All-Cause Readmissions (PCR) SUMMARY OF CHANGES TO HEDIS 2015 Removed the former step 4 to exclude acute inpatient stays with a discharge date in the 30 days prior to the Index Admission Date. Added step 5 (required exclusions) to exclude acute inpatient discharges followed by a planned readmission within 30 days. Removed gender strata from reporting table PCR-A-2/3 and PCR-B-3. Rates will be reported for both genders combined by age strata. Description For members 18 years of age and older, the number of acute inpatient stays during the measurement year that were followed by an unplanned acute readmission for any diagnosis within 30 days and the predicted probability of an acute readmission. Data are reported in the following categories: 1. Count of Index Hospital Stays (IHS) (denominator). 2. Count of 30-Day Readmissions (numerator). 3. Average Adjusted Probability of Readmission. Note: For commercial, only members years of age are reported. Definitions HIS Index Admission Date Index Discharge Date Index Readmission Stay Index Readmission Date Index hospital stay. An acute inpatient stay with a discharge on or between January 1 and December 1 of the measurement year. Exclude stays that meet the exclusion criteria in the denominator section. The IHS admission date. The IHS discharge date. The index discharge date must occur on or between January 1 and December 1 of the measurement year. An acute inpatient stay for any diagnosis with an admission date within 30 days of a previous Index Discharge Date. The admission date associated with the Index Readmission Stay. Planned hospital stay A hospital stay is considered planned if it meets criteria as described in step 5 (required exclusions) of the Eligible Population. Classification period 365 days prior to and including an Index Discharge Date.

3 Draft Document for HEDIS 2015 Public Comment Obsolete After March 19, Risk Adjustment Tables Table Table Description HCC-Surg Surgery codes for Risk Adjustment Determination PCR-DischCC Discharge Clinical Condition category codes for Risk Adjustment Determination CC-Comorbid Comorbid Clinical Condition category codes for Risk Adjustment Determination step 2 HCC Rank HCC rankings for Risk Adjustment Determination step 3 HCC-Comb Combination HCCs for Risk Adjustment Determination step 5 PCR-MA-DischCC-Weight- Under65 MA and SNP primary discharge weights for Risk Adjustment Weighting step 2 for ages under 65 PCR-MA-DischCC-Weight-65plus MA and SNP primary discharge weights for Risk Adjustment Weighting step 2 for ages 65 and older PCR-Comm-DischCC-Weight Commercial primary discharge weights for Risk Adjustment Weighting step 2 PCR-MA-ComorbHCC-Weight- MA and SNP comorbidity weights for Risk Adjustment Weighting step 3 for ages under 65 Under65 PCR-MA-ComorbHCC-Weight- 65plus MA and SNP comorbidity weights for Risk Adjustment Weighting step 3 for ages 65 and older PCR-Comm-ComorbHCC-Weight Commercial comorbidity weights for Risk Adjustment Weighting step 3 PCR-MA-OtherWeights-Under65 MA and SNP base risk, surgery, age and gender weights for Risk Adjustment Weighting steps 1, 4, 5 for ages under 65 PCR-MA-OtherWeights-65plus MA and SNP base risk, surgery, age and gender weights for Risk Adjustment Weighting steps 1, 4, 5 for ages 65 and older PCR-Comm-OtherWeights Commercial base risk, surgery, age and gender weights for Risk Adjustment Weighting steps 1, 4, 5 Note: The risk adjustment tables will be released on November 1, 2013, and posted to Eligible Population Product line Ages Commercial, Medicare (report each product line separately). For commercial, ages as of the Index Discharge Date. For Medicare, ages 18 and older as of the Index Discharge Date. Continuous enrollment Allowable gap Anchor date Benefit Event/ diagnosis 365 days prior to the Index Discharge Date through 30 days after the Index Discharge Date. No more than one gap in enrollment of up to 45 days during the 365 days prior to the Index Discharge Date and no gap during the 30 days following the Index Discharge date. Index Discharge Date. Medical. An acute inpatient discharge on or between January 1 and December 1 of the measurement year. The denominator for this measure is based on discharges, not members. Include all acute inpatient discharges for members who had one or more discharges on or between January 1 and December 1 of the measurement year. The organization should follow the steps below to identify acute inpatient stays.

4 Draft Document for HEDIS 2015 Public Comment Obsolete After March 19, Administrative Specification Denominator Step 1 The eligible population. Identify all acute inpatient stays with a discharge date on or between January 1 and December 1 of the measurement year. Include acute admissions to behavioral healthcare facilities. Exclude nonacute inpatient rehabilitation services, including nonacute inpatient stays at rehabilitation facilities. Step 2 Step 3 Step 4 Step 45 Step 5 Acute-to-acute transfers: Keep the original admission date as the Index Admission Date, but use the transfer s discharge date as the Index Discharge Date. Exclude hospital stays where the Index Admission Date is the same as the Index Discharge Date. Exclude any acute inpatient stay with a discharge date in the 30 days prior to the Index Admission Date. Exclude stays for the following reasons: Acute inpatient discharges for death. Acute inpatient discharge with a principal diagnosis of pregnancy (Pregnancy Value Set). Acute inpatient discharge with a principal diagnosis of a condition originating in the perinatal period (Perinatal Conditions Value Set). For all acute inpatient discharges identified using steps 1 4, determine if there was a planned readmission within 30 days using all acute inpatient stays. Exclude any acute inpatient discharge as an Index Hospital Stay if the admission date of the first readmission is within 30 days and includes any of the following. A principal diagnosis of maintenance chemotherapy (Chemotherapy Value Set). A principal diagnosis of rehabilitation (Rehabilitation Value Set). An organ transplant (Kidney Transplant Value Set, Bone Marrow Transplant Value Set; Organ Transplant Other Than Kidney Value Set). A potentially planned procedure (Potentially Planned Procedure Value Set) without a principal acute diagnosis (Acute Condition Value Set). Example 1 For a member with the following acute inpatient stays, exclude stay 1. Stay 1 (January 30 February 1 of the measurement year): Acute inpatient discharge with a principal diagnosis of COPD. Stay 2 (February 5 7 of the measurement year): Acute inpatient discharge with a principal diagnosis of chemotherapy. Example 2 For a member with the following acute inpatient stays, exclude stays 2 and 3 in the following scenario. Stay 1 (January of the measurement year): Acute inpatient discharge with a principal diagnosis of diabetes Stay 2 (January 30 February 1 of the measurement year): Acute inpatient discharge with a principal diagnosis of COPD. Stay 3 (February 5 7 of the measurement year): Acute inpatient discharge with an organ transplant.

5 Draft Document for HEDIS 2015 Public Comment Obsolete After March 19, Step 56 Step 67 Stay 4 (February of the measurement year): Acute inpatient discharge with a principal diagnosis of rehabilitation. Calculate continuous enrollment. Assign each acute inpatient stay to one age and gender category. Refer to Tables PCR-A-2/3 and Table PCR-B-3. Risk Adjustment Determination For each IHS, use the following steps to identify risk adjustment categories based on presence of surgeries, discharge condition, comorbidity, age and gender. Surgeries Discharge Condition Determine if the member underwent surgery during the inpatient stay. Download the list of codes from the NCQA Web site (Table HCC-Surg) and use it to identify surgeries. Consider an IHS to include a surgery if at least one procedure code in Table HCC-Surg is present from any provider between the admission and discharge dates. Assign a discharge Clinical Condition (CC) category code to the IHS based on its primary discharge diagnosis, using Table PCR-DischCC. For acute-to-acute transfers, use the transfer s primary discharge diagnosis. Exclude diagnoses that cannot be mapped to Table PCR-DischCC. Comorbidities Step 1 Step 2 Identify all diagnoses for encounters during the classification period. Include the following when identifying encounters: Outpatient visits (Outpatient Value Set). Observation visits (Observation Value Set). Nonacute inpatient encounters (Nonacute Inpatient Value Set). Acute inpatient encounters (Acute Inpatient Value Set). ED visits (ED Value Set). Exclude the primary discharge diagnosis on the IHS. Assign each diagnosis to one comorbid Clinical Condition (CC) category using Table CC Comorbid. Exclude all diagnoses that cannot be assigned to a comorbid CC category. For members with no qualifying diagnoses from face-to-face encounters, skip to the Risk Adjustment Weighting section. All digits must match exactly when mapping diagnosis codes to the comorbid CCs. Step 3 Determine HCCs for each comorbid CC identified. Refer to Table HCC Rank. For each stay s comorbid CC list, match the comorbid CC code to the comorbid CC code in the table, and assign: The ranking group. The rank. The HCC.

6 Draft Document for HEDIS 2015 Public Comment Obsolete After March 19, For comorbid CCs that do not match to Table HCC Rank, use the comorbid CC as the HCC and assign a rank of 1. Note: One comorbid CC can map to multiple HCCs; each HCC can have one or more comorbid CCs. Step 4 Assess each ranking group separately and select only the highest ranked HCC in each ranking group using the Rank column (1 is the highest rank possible). Drop all other HCCs in each ranking group, and de-duplicate the HCC list if necessary. Example Assume a stay with the following comorbid CCs: CC-15, CC-19 and CC-80 (assume no other CCs). CC-80 does not have a map to the ranking table and becomes HCC-80. HCC-15 is part of Ranking Group 1 and HCC-19 is part of Ranking Groups Diabetes 1 Diabetes 4. Because CC-15 is ranked higher than CC-19 in Ranking Group Diabetes 1, the comorbidity is assigned as HCC-15 for Ranking Group 1. Because CC-19 is ranked higher in Ranking Groups Diabetes 2 4, the comorbidity is assigned as HCC-19 for these ranking groups. The final comorbidities for this discharge are HCC-15, HCC-19 and HCC-80. Example: Table HCC Rank Ranking Group CC Description Rank HCC NA CC-80 Congestive Heart Failure NyA HyCC-80 Diabetes 1 CC-15 Diabetes With Renal or Peripheral Circulatory 1 HCC-15 Manifestation CC-16 Diabetes With Neurologic or Other Specified 2 HCC-16 Manifestation CC-17 Diabetes With Acute Complications 3 HCC-17 CC-18 Diabetes With Ophthalmologic or Unspecified 4 HCC-18 Manifestation CC-19 Diabetes Without Complications 5 HCC-19 Diabetes 2 CC-16 Diabetes With Neurologic or Other Specified 1 HCC-16 Manifestation CC-17 Diabetes With Acute Complications 2 HCC-17 CC-18 Diabetes With Ophthalmologic or Unspecified 3 HCC-18 Manifestation CC-19 Diabetes Without Complication 4 HCC-19 Diabetes 3 CC-17 Diabetes With Acute Complications 1 HCC-17 CC-18 Diabetes With Ophthalmologic or Unspecified 2 HCC-18 Manifestation CC-19 Diabetes Without Complication 3 HCC-19 CC-18 Diabetes With Ophthalmologic or Unspecified 1 HCC-18 Diabetes 4 Manifestation CC-19 Diabetes Without Complication 2 HCC-19

7 Draft Document for HEDIS 2015 Public Comment Obsolete After March 19, Step 5 Identify combination HCCs listed in Table HCC Comb. Some combinations suggest a greater amount of risk when observed together. For example, when diabetes and CHF are present, an increased amount of risk is evident. Additional HCCs are selected to account for these relationships. Compare each stay s list of unique HCCs to those in the HCC column in Table HCC Comb and assign any additional HCC conditions. For fully nested combinations (e.g., the diabetes/chf combination is nested in the diabetes/ CHF/renal combination), use only the more comprehensive pattern. In this example, only the diabetes/chf/renal combination is counted. For overlapping combinations (e.g., the CHF, COPD combination overlaps the CHR/renal/ diabetes combination), use both sets of combinations. In this example, both CHF/COPD and CHF/renal/diabetes combinations are counted. Based on the combinations, a member can have none, one or more of these added HCCs. Example For a stay with comorbidities HCC-15, HCC-19 and HCC-80 (assume no other HCCs), assign HCC-901 in addition to HCC-15, HCC-19 and HCC-80. This does not replace HCC-15, HCC- 19 or HCC-80. Example: Table HCC Comb Combination: Diabetes and CHF Comorbid HCC Comorbid HCC Comorbid HCC Combination HCC HCC-15 HCC-80 NA HCC-901 HCC-16 HCC-80 NA HCC-901 HCC-17 HCC-80 NA HCC-901 HCC-18 HCC-80 NA HCC-901 HCC-19 HCC-80 NA HCC-901 Risk Adjustment Weighting For each IHS, use the following steps to identify risk adjustment weights based on presence of surgeries, discharge condition, comorbidity, age and gender. Note: The final weights table will be released on November 1, Step 1 Step 2 For each IHS with a surgery, link the surgery weight. For Medicare product lines ages 18 64: Use Table PCR-MA-OtherWeights-Under65. For Medicare product lines ages 65 and older: Use Table PCR-MA-OtherWeights- 65plus. For commercial product lines: Use Table PCR-Comm-OtherWeights. For each IHS with a discharge CC Category, link the primary discharge weights. For Medicare product lines ages 18-64: Use Table PCR-MA-DischCC-Weight-Under65. For Medicare product lines ages 65 and older: Use Table PCR-MA-DischCC-Weight- 65plus. For commercial product lines: Use Table PCR-Comm-DischCC-Weight.

8 Draft Document for HEDIS 2015 Public Comment Obsolete After March 19, Step 3 For each IHS with a comorbidity HCC Category, link the weights. For Medicare product lines ages 18 64: Use Table PCR-MA-ComorbHCC-Weight- Under65. For Medicare product lines ages 65 and older: Use Table PCR-MA-ComorbHCC-Weight- 65plus. For commercial product lines: Use Table PCR-Comm-ComorbHCC-Weight. Step 4 Link the age and gender weights for each IHS. For Medicare product lines ages 18 64: Use Table PCR-MA-OtherWeights-Under65. For Medicare product lines ages 65 and older: Use Table PCR-MA-OtherWeights-65plus. For commercial product lines: Use Table PCR-Comm-OtherWeights. Step 5 Identify the base risk weight. For Medicare product lines ages 18 64: Use Table PCR-MA-OtherWeights-Under65. For Medicare product lines ages 65 and older: Use Table PCR-MA-OtherWeights-65plus. For commercial product lines: Use Table PCR-Comm-OtherWeights to determine the base risk weight. Step 6 Step 7 Sum all weights associated with the IHS (i.e., presence of surgery, primary discharge diagnosis, comorbidities, age, gender and base risk weight). Use the formula below to calculate the adjusted probability of a readmission based on the sum of the weights for each IHS. Adjusted probability of readmission = WeightsForIHS WeightsForIHS OR Adjusted probability of readmission = [exp (sum of weights for IHS )] / [ 1 + exp (sum of weights for IHS) ] Note: Exp refers to the exponential or antilog function. Step 8 Use the formula below and the adjusted probability of readmission calculated in step 7 to calculate the variance for each IHS. Variance = Adjusted probability of readmission x (1 Adjusted probability of readmission) Example: If the adjusted probability of readmission is for an IHS, then the variance for this IHS is x = Note: The variance is calculated at the IHS level. Organizations must sum the variances for each age/gender and total category when populating the Total Variance cells in the reporting tables.

9 Draft Document for HEDIS 2015 Public Comment Obsolete After March 19, Sample Table: PCR Risk Adjustment Weighting Member ID* Admiss. Counter Base Risk Weight Age Gender Age and Gender Weight Surgical Weight ICD-9 Diagnosis Code Discharge CC Female HCC-PCR Category Weight Category Weight Sum of Weights Adjusted Probability Variance Male NA NA NA Male NA *Each Member ID field with a value represents a unique IHS Numerator Step 1 Step 2 Step 3 Step 4 At least one acute readmission for any diagnosis within 30 days of the Index Discharge Date. Identify all acute inpatient stays with an admission date on or between January 2 and December 31 of the measurement year. Acute-to-acute transfers: Keep the original admission date as the Index Admission Date, but use the transfer s discharge date as the Index Discharge Date. Exclude acute inpatient hospital discharges with a principal diagnosis of pregnancy (Pregnancy Value Set) or a principal diagnosis for a condition originating in the perinatal period (Perinatal Conditions Value Set). For each IHS, determine if any of the acute inpatient stays have an admission date within 30 days after the Index Discharge Date.

10 Draft Document for HEDIS 2015 Public Comment Obsolete After March 19, Reporting: Denominator Count the number of IHS for each age, gender and total combination and enter these values into the reporting table. Reporting: Risk Adjustment Step 1 Calculate the average adjusted probability for each IHS for each age, gender and total combinations and the overall total. Organizations must calculate the probability of readmission for each hospital stay within the applicable age and gender group to calculate the average (which is reported to NCQA). For the total age/gender category, the probability of readmission for all hospital stays in the age/gender categories must be averaged together; organizations cannot take the average of the average adjusted probabilities reported for each age/gender. Step 2 Example Round to four decimal places using the.5 rule and enter these values into the reporting table. Note: Do not take the average of the cells in the reporting table. For the age category: Identify all IHS by year-old males and calculate the average adjusted probability. Identify all IHS by year-old females and calculate the average adjusted probability. Identify all IHS by all year-olds and calculate the average adjusted probability. Repeat for each subsequent group. Step 3 Step 4 Calculate the total (sum) variance for each age, gender and total combinations and the overall total. Round to four decimal places using the.5 rule and enter these values into the reporting table. Reporting: Numerator Count the number of IHS with a readmission within 30 days for each age, gender and total combination and enter these values into the reporting table. Note Organizations may not use Risk Assessment Protocols to supplement diagnoses for calculation of the risk adjustment scores for this measure. The PCR measurement model was developed and tested using only claims-based diagnoses and diagnoses from additional data sources would affect the validity of the models as they are current implemented in the specification.

11 Draft Document for HEDIS 2015 Public Comment Obsolete After March 19, Table PCR-A-2/3: Plan All-Cause Readmissions Rates by Age, Gender and Risk Adjustment Age Total Count of Index Stays (Denominator) Count of 30- Day Readmit (Numerator) Observed Readmit (Num/Den) Average Adjusted Probability Sex Total Variance Male Female Total: Male Female Total: Male Female Total: O/E Ratio (Obs. Readmit/ Avg. Adjusted Probability) Lower Confidence Interval (O/E Ratio) Upper Confidence Interval (O/E Ratio) Male Female Total: Table PCR-B-3: Plan All-Cause Readmissions Rates by Age, Gender and Risk Adjustment Age Total Count of Index Stays (Denominator) Count of 30- Day Readmit (Numerator) Observed Readmit (Num/Den) Average Adjusted Probability Sex Total Variance Male Female Total: Male Female Total: Male Female Total: O/E Ratio (Obs. Readmit/ Avg. Adjusted Probability) Lower Confidence Interval (O/E Ratio) Upper Confidence Interval (O/E Ratio) Male Female Total:

12 Draft Document for HEDIS 2015 Public Comment Obsolete After March 19, Plan All Cause Readmission Measure Work-Up Measure Description For members 18 years of age and older, the number of acute inpatient stays during the measurement year that were followed by an unplanned acute readmission for any diagnosis within 30 days and the predicted probability of an acute readmission. Data are reported in the following categories: 1. Count of Index Hospital Stays (IHS) (denominator). 2. Count of 30-Day Readmissions (numerator). 3. Average Adjusted Probability of Readmission. A final rate is reported as ratio of the observed rate of readmission over the expected rate of readmission based on the age, gender, discharge diagnosis and comorbid conditions of the discharged population. Topic Overview Importance and Prevalence Health importance Discharge from the hospital is a critical transition point in a patient s care. Incomplete handoffs at discharge and poor care coordination can lead to adverse events for patients and avoidable rehospitalization. Hospital readmissions may indicate poor care or missed opportunities to better coordinate care (MedPAC, 2007). Hospital readmission is associated with longer lengths of stay and higher mortality for patients. A Dartmouth- Hitchcock Medical Center found that hospital mortality was significantly higher for readmitted patients in the intensive care unit (ICU). This retrospective cohort study also showed that a hospital s length of stay was higher for patients that were readmitted (Cook, 2006). Similarly, a United Kingdom study showed that mortality was significantly higher for patients readmitted to the ICU (death rates were 1.5 to almost 10 times higher among readmission patients). Even after risk adjusting for disease and severity category, studies found that the odds of death remained six and seven times higher among readmitted patients (Rosenberg, 2000). Recent studies also suggest that older patients tend to experience substantial cognitive decline following hospitalization, even after controlling for severity of illness and cognitive decline that took place before hospital admission (Rockwood, 2012; Wilson et al., 2012). Prevalence A recent MEDPAC report to Congress stated that despite a recent slight decline in readmission rates, 12.3 percent of all 2011 Medicare admissions were followed by a potentially preventable readmission. Readmission rates ranged from 9.9 percent for the hospital at the 10th percentile of the distribution to 15.3 percent at the 90th percentile (MEDPAC, 2013). Nationally representative data collected from 2011 and 2012 and reported by the Department of Health and Human Services on Hospital Compare ( shows an average hospital readmission rate of 16.0 percent. Historically, rates of hospital readmission for Medicare patients were as high as 19 percent within 30 days of discharge (Jencks et al., 2009). Hospital readmissions are commonly related to conditions such as congestive heart failure, acute myocardial infarction, chronic obstructive pulmonary disease and pneumonia (MEDPAC, 2013). Nationally representative data collected from 2009

13 Draft Document for HEDIS 2015 Public Comment Obsolete After March 19, and reported by the Department of Health and Human Services on Hospital Compare showed that heart failure patients had the highest rate of hospital readmission (23 percent), followed by heart attack patients (18.3 percent), and pneumonia patients (17.6 percent) ( Many experts, including the Institute for Healthcare Improvement, concur that congestive heart failure as a reason for hospital readmission can be considered potentially preventable (MedPAC, 2007; IHI, 2004). For Medicare patients with congestive heart failure, 15-day readmission rates average 12.5 percent. Roughly 20 percent of hospitals that treat patients who have congestive heart failure have inpatient readmission rates more than 4 percent higher than expected (MedPAC, 2007). Financial importance and cost effectiveness Unplanned hospitalizations are not only a burden for patients, they also have huge financial costs. Studies have found that patients who receive post-discharge interventions have reduced hospital readmission and less total health care costs than those who do not receive such interventions (Constantino et al., 2013; Harrison et al., 2011). According to the Medicare Provider Analysis and Review file data, in 2005, 5.2 percent of Medicare patient readmissions within 7 days of discharge were considered potentially preventable; 8.8 percent of readmissions within 15 days were potentially preventable; and 13.3 percent of readmissions within 30 days were potentially preventable. This equates to $5 billion, $8 billion and $12 billion dollars, respectively, for potentially preventable readmissions. The average Medicare payment for a potentially preventable readmission totaled approximately $7,200 (MedPac, 2007). In 2005, the 30-day hospital readmission rates for Medicare patients ranged from 14 percent 22 percent. If readmission rates were lowered to the levels achieved by the top-performing regions, Medicare would save $1.9 billion annually (Commonwealth Fund, 2006). Supporting Evidence for Reducing Unplanned Readmissions This measure is not based on clinical guidelines, but is supported by research and policy interest in reducing costs and improving patient care. Research shows that specific hospital-based initiatives to improve communication with beneficiaries and their caregivers, coordination of care after discharge and improving the quality of care during the initial admission can avert many readmissions (MedPac, 2007). Measuring readmissions must be clearly defined to distinguish between measures of all readmissions that may not correlate to the quality of care provided or may not be preventable, and measures that focus on potentially preventable admissions. A large number of interventional and observational studies, including a number of randomized controlled trials, have explored methods for preventing readmissions. In a meta-review of published systematic reviews of the effect of clinical interventions on hospital readmission rates, Benbassat and Targin (2013) found that disease management programs in the community significantly reduced readmission rates in patients with heart failure, coronary heart disease and bronchial asthma. They found less consistent support for in-hospital interventions: some studies and meta-analyses showed in-hospital interventions lead to reduced readmissions; others found limited evidence of reduced readmissions post-intervention. Despite inconsistent findings in the literature, there is a strong and growing consensus, as evidenced by examples of health plan successes detailed below, that a substantial subset of readmissions are avoidable and that more effective care can reduce their number.

14 Draft Document for HEDIS 2015 Public Comment Obsolete After March 19, Health plan role Gaps in care Health plans can play a critical role in improving the quality of care transitions and reducing the rate of hospital readmissions. Numerous health plans have established quality improvement efforts that have resulted in a reduction in readmission rates. Below are a few examples of successful health plan interventions to reduce readmissions. Kaiser Permanente Northwest Region reduced readmission rates from December 2010 November 2012, from 12.8 percent 11 percent. The change in O/E ratio was The plan implemented a multidimensional intervention that included risk stratification, standardized discharge summary, medication reconciliation, postdischarge phone calls, timely follow-up with a primary care physician, a special transition phone number, palliative care consult and complex-case conferences (Tuso et al., 2013). In California and Massachusetts Medicare Advantage plans, a model of pharmacist in-home visits with patients after discharge and comprehensive medication management resulted in up to a 30 percent reduction in hospital readmission rates (Novac et al., 2012). In a retrospective study of more than 100,000 Medicare Advantage beneficiaries, Costantino et al. (2013) found that implementation of a post-discharge telephone intervention reduced readmissions, compared with a control group (9.3 percent and 11.5 percent, respectively; p<0.0001). As a group, overall cost savings were $499,458 for members who received the intervention, with $13,964,773 in savings to the health plan. A similar randomized controlled trial of a post-discharge telephone intervention in a commercial health insurance population found a 22 percent reduction in readmissions for the treatment group, compared with the control group (Melton et al., 2012). Recent data from the HEDIS Health Plan measure showed average rates of hospital readmission in 2012 to be 15.3 (SD=3.5) for year-olds in Medicare Advantage HMO plans and 15.4 percent (SD=3.5) for year-olds in Medicare Advantage PPO plans. The average hospital readmission rate for 65 and older in Medicare Advantage HMO was 14.0 percent (SD=3.0) and was 13.0 percent (SD=2.8) for those in Medicare Advantage PPO. Medicare Advantage plans serving had the most variation in performance, with an 8.5 percent difference in rates between HMO plans at the 10th and 90th percentiles, and a 7.7 percent difference for PPO plans at the 10th and 90th percentiles. Commercial plans had lower rates of hospital readmissions and less variation between the high and low performers. Commercial HMO plans averaged 9.0 percent (SD=4.8) and PPO plans averaged 8.2 percent (SD=1.1). These data suggest there is significant room for improvement, particularly for Medicare plans. International comparisons of hospital readmission rates also suggest a need for improvement in the United States. In a study comparing readmission rates in 17 different countries, Kociol et al. (2012) found that 30-day post-discharge readmission rates for patients with myocardial infarction were 68 percent higher in the United States than the average for European countries from 2006 through Jencks et al. (2009) found great variation in hospital readmission rates across different states and hospitals, suggesting there is room for improvement. Currently, use of recommended practices to reduce readmissions varies widely across hospitals in the United States (Bradley et al., 2012).

15 Draft Document for HEDIS 2015 Public Comment Obsolete After March 19, Health care disparities Research suggests that racial and ethnic minorities are more likely to experience hospital readmissions than Whites. In a study of hospital readmissions within 6 months, researchers at the Agency for Healthcare Research and Quality found that older African Americans, Native Americans and Hispanics had significantly higher rates of readmission than older White adults (Friedman and Basu, 2004). In evaluations of 180-day readmission rates for patients with diabetes, Jiang et al. (2005) found readmission rates varied by race/ethnicity and type of insurance, as follows: White African American Hispanic Commercial 21.4% 20.2% 23.1% Medicaid 32.5% 33.0% 34.2% Medicare 27.9% 30.7% 34.0% Rathore et al. (2003) found that Black Medicare patients had higher rates of readmission following heart failure treatment than White Medicare patients. In a more recent study of 2008 Medicare hospital readmission data, McHugh et al. (2010) found that Black Medicare beneficiaries with heart failure, acute myocardial infarction and pneumonia were more likely than Whites to be readmitted following an initial hospitalization, while Hispanic beneficiaries had significantly higher odds of readmission for acute myocardial infarction. Lower socioeconomic status has been found to be a risk factor for hospital readmission. In a retrospective analysis of medical record data from a national sample of Medicare beneficiaries hospitalized with heart failure (n=25,086), researchers found that lower socioeconomic status of patients was correlated with higher hospital readmission rates (Rathore et al., 2006). On a facility level, hospitals that serve a higher share of low-income patients are 30 percent more likely to have 30-day hospital readmission rates above the national average than hospitals that serve a lower share (Berenson and Shih, 2012). Recent analysis by the Medicare Payment Advisory Commission also suggests that hospitals readmission rates are positively correlated with their low-income patient share. In 2013, the Commission s analysis found that a hospital s share of low-income patients (defined as Medicare patients receiving Social Security income) was a stronger and more consistent predictor of readmissions than was patient race (MEDPAC, 2013). References Benbassat, J., and M.I. Taragin The effect of clinical interventions on hospital readmissions: a metareview of published meta-analyses. Israel Journal of Health Policy Research, 2(1):1 15. Bradley, E.H., L. Curry, L.I. Horwitz, H. Sipsma, J.W. Thompson, M. Elma,... and H.M. Krumholz Contemporary Evidence About Hospital Strategies for Reducing 30-Day Readmissions: A National Study. Journal of the American College of Cardiology, 60(7): Berenson, J. and A. Shih December Higher Readmissions at Safety-Net Hospitals and Potential Policy Solutions, The Commonwealth Fund. Costantino, M.E., B. Frey, B. Hall, and P. Painter The Influence of a Postdischarge Intervention on Reducing Hospital Readmissions in a Medicare Population. Population Health Management, 16(5). Friedman, B., J. Basu The rate and cost of hospital readmissions for preventable conditions. Medical Care Research and Review. 61: Harrison, P.L., P.A. Hara, J.E. Pope, M.C. Young, and E.Y. Rula The impact of postdischarge telephonic follow-up on hospital readmissions. Population Health Management. 14(1): Institute for Healthcare Improvement Reducing readmissions for heart failure patients: Hackensack University Medical Center. Jencks, S.F., M.V. Williams, and E.A. Coleman Rehospitalizations among patients in the icare fee-forservice program. New England Journal of Medicine. 360(14):

16 Draft Document for HEDIS 2015 Public Comment Obsolete After March 19, Jiang, H.J., R. Andrews, D. Stryer, and B. Friedman Racial/Ethnic Disparities in Potentially Preventable Readmissions: The Case of Diabetes. American Journal of Public Health. 95(9): Kociol, R.D., R.D. Lopes, R. Clare, L. Thomas, R.H. Mehta, P. Kaul,... and M.R. Patel International variation in and factors associated with hospital readmission after myocardial infarction. JAMA: the journal of the American Medical Association. 307(1): McHugh, M.D., J.M.B. Carthon, and X.L. Kang Medicare readmissions policies and racial and ethnic health disparities: a cautionary tale. Policy, Politics, & Nursing Practice. 11(4): Medicare Payment Advisory Commission. June Report to the Congress: Promoting Greater Efficiency in Medicare. (October 13, 2008) Medicare Payment Advisory Commission. June Refining the Hospital Readmissions Reduction Program, Chapter 4 in Report to Congress: Medicare and the Health Care Delivery System (Washington, D.C.: MedPAC). Melton, et al Prioritized Post-Discharge Telephonic Outreach Reduces Hospital Readmissions for Select High-Risk Patients. American Journal of Managed Care. 18(12). Novak, C.J., S. Hastanan, M. Moradi, and D.F. Terry Reducing unnecessary hospital readmissions: the pharmacist's role in care transitions. The Consultant Pharmacist, 27(3):174 9 The Commonwealth Fund. September The Commonwealth Fund Commission on a High Performance Health System, Why Not the Best? Results from a National Scorecard on U.S. Health System Performance. The Commonwealth Fund The Commonwealth Fund Commission on a Case Study: Reducing Hospital Readmissions Among Heart Failure Patients at Catholic Healthcare Partners. Rathore, S.S., J.A.M. Foody, Y. Wang, G.L. Smith, J. Herrin, F.A. Masoudi, H.M. Krumholz Race, quality of care, and outcomes of elderly patients hospitalized with heart failure. Journal of the American Medical Association. 289:2517. Rathore, S.S., F.A. Masoudi, Y. Wang, J.P. Curtis, J.M. Foody, E.P. Havranek, and H.M. Krumholz Socioeconomic status, treatment, and outcomes among elderly patients hospitalized with heart failure: findings from the national heart failure project. American Heart Journal. 152(2): Rockwood, K Hospitalization and effects on cognition. Neurology. 78(13):e86 7. Rosenberg, A.L., and C. Watts Patients Readmitted to ICUs. A systematic review of risk factors and outcomes. Critical Care Reviews. Chest. 118: Tuso, P., D.N. Huynh, D.P.P.D. Lynn Garofalo, G. Lindsay, M.H.A. Brandy Florence, J. Jones,... and M.H. Kanter The Readmission Reduction Program of Kaiser Permanente Southern California Knowledge Transfer and Performance Improvement. The Permanente Journal. 17(3). Wilson, R.S., L.E. Hebert, P.A. Scherr, et al Cognitive decline after hospitalization in a community population of older persons. Neurology. 78(13):950 6.

17 Draft Document for HEDIS 2015 Public Comment Obsolete After March 19, HEDIS Health Plan Performance Rates: Plan All-Cause Readmission (PCR) Note: Lower rates signify better performance PRODUCT LINE: COMMERCIAL Table 1. HEDIS PCR Measure Performance Observed Rate of Readmission Commercial HMO Plans Year Age Group Total Number of Plans Plans Able to Report (%) Average Standard Deviation 10th 25th 50th 75th 90th (93.2) (95.9) Table 2. HEDIS PCR Measure Performance Observed Rate of Readmission Commercial PPO Plans Age Total Number Plans Able to Standard 10th 25th 50th 75th 90th Year Group of Plans Report (%) Average Deviation (99.0) (98.5) Table 3. HEDIS PCR Measure Performance O/E Ratio Commercial HMO Plans Age Total Number Plans Able to Standard 10th 25th 50th 75th 90th Year Group of Plans Report (%) Average Deviation (93.2) (95.9) Table 4. HEDIS PCR Measure Performance O/E Ratio Commercial PPO Plans Age Total Number Plans Able to Standard 10th 25th 50th 75th 90th Year Group of Plans Report (%) Average Deviation (99.0) (98.5)

18 Draft Document for HEDIS 2015 Public Comment Obsolete After March 19, PRODUCT LINE: MEDICARE Table 1. HEDIS PCR Measure Performance Observed Rate of Readmission Medicare HMO Plans Year Age Total Number Plans Able to Standard 10th 25th 50th 75th 90th Group of Plans Report (%) Average Deviation (88.0) (96.2) (80.3) (89.2) Table 2. HEDIS PCR Measure Performance Observed Rate of Readmission Medicare PPO Plans Year Age Group Total Number of Plans Plans Able to Report (%) Average Standard Deviation 10th 25th 50th 75th 90th (85.6) (98.6) ( (95.5) Table 3. HEDIS PCR Measure Performance O/E Ratio Medicare HMO Plans Year Age Group Total Number of Plans Plans Able to Report (%) Average Standard Deviation 10th 25th 50th 75th 90th (88.0) (96.2) (80.3) (89.2) Table 4. HEDIS PCR Measure Performance O/E Ratio Medicare PPO Plans Year Age Group Total Number of Plans Plans Able to Report (%) Average Standard Deviation 10th 25th 50th 75th 90th (85.6) (98.6) (81.3) (95.5)