Health Information Technology and Management

Size: px

Start display at page:

Download "Health Information Technology and Management"

Richard Barrett
6 years ago
Views:

1 Health Information Technology and Management CHAPTER 11 Health Statistics, Research, and Quality Improvement

2 Pretest (True/False) Children s asthma care is an example of one of the core measure sets for which hospitals are required to collect and transmit data to the Joint Commission. When a hospital does not have enough cases to meet the minimum needed for data sampling, then sampling is not done. The formula to calculate a ratio is x/y = ratio. Rates are used to measure events occurring over a period of time.

3 Indexes Forms of Secondary Health Records Either separate files or pointers to data within primary health records Registries Usually separate databases created to track specific types of data like cancer tumors, implanted devices, or childhood immunizations Custom data sets Used for reporting performance (HEDIS, NHQM) (Secondary data examples: health insurance claim, MPI, and aggregate data) (HEDIS Health Plan Employer Data and Information Set) (NHQM National Hospital Quality Measures)

4 Indexes Created manually in a paper system; automatically in EHR system Organized by various categories, such as disease, attending physician, surgeon, procedures, discharge status, patient s age or zip code, and so on Internal uses Permit healthcare organizations to locate, count, and analyze data for quality and process improvement External uses Allow quick identification and sorting of records for external reporting Allow automatic identification of records for abstracting for both internal or external registries

5 Registries Separate databases created to track specific types of data Available to either internal or external users Hospitals use to improve performance or processes, or to satisfy accreditation (Internal) May be required by outside sources for data reporting (External) Examples: trauma, cancer tumor, implanted device, childhood immunization registries

6 Example: Cancer Registries Facility-based cancer registrar enters data about cases by abstracting it from the health records of patients diagnosed with some form of cancer Identifies cases using disease index, discharge reports, pathology reports, and/or patient registration Internal use Facility quality assessment and research, and measure success of various treatment modalities External use Aggregated and reported to state and national cancer registries Used to identify trends and changes in the incidence and survival rate

7 Index Versus Registry Index Points to medical record containing one or more fields to be reported or studied Example: disease index includes all patients Registry Separate database into which certain data elements have been imported or manually entered Example: hospital trauma registry would include entries added by selecting cases with certain diagnosis codes

8 Custom Data Sets: HEDIS (Health Plan Employer Data and Information Set) Created by NCQA (National Committee for Quality Assurance) as a tool to compare quality of care patients receive under various health plans Consists of 71 measures across eight domains of care Allows employers to use results of NCQA reports to select best plan for employees NCQA has accreditation program for health plans NCQA collects HEDIS data directly from managed care HMO and PPO organizations Researchers may use HEDIS data to study trends Does not contain PHI

9 Custom Data Sets: ORYX Helps integrate outcomes and other performance measurement data into accreditation process Supports healthcare organizations in internal quality improvement efforts Standardized with CMS, allowing facility to collect and report same data set for both Joint Commission and CMS initiatives Called National Hospital Quality Measures (NHQM)

10 ORYX: Available Core Measure Sets Acute myocardial infarction (AMI) heart attack Heart failure (HF) Pneumonia (PN) Pregnancy and related conditions (PR) Hospital-based inpatient psychiatric services (HBIPS) Children s asthma care (CAC) Surgical Care Improvement Project (SCIP) Hospital outpatient program quality measures (HOP)

11 Data Analysis Sampling Algorithms

12 Data Sampling Selecting a subset of data records using an algorithm Data analysis includes applying mathematical formulas to produce statistical studies Sample size refers to number of cases necessary to make sample meaningful Data must also include type of cases that apply to measure and exclude those that are not applicable Joint Commission and CMS have determined minimum number of cases that would produce statistically valid samples for each measure set; (for example, AMI must have at least 35 cases.)

13 Figure 11-4 Steps for sampling NHQM data.

14 Algorithms Predefined set of rules that helps break down complex processes into simple, repetitive steps Used to process data to arrive at desired result Used to select initial patient population in data sampling for the measure set

15 Figure 11-5 Flow of algorithm for acute myocardial infarction measure set.

16 Healthcare Statistics Ratios Proportions Rates Measures of Central Tendency (mean, median, mode) Measures of Variability (range, variance, standard deviation)

17 Ratios Can be used to show the relationship between two different things Usually written as 2 numbers separated by colon, such as 9:1 Formula: x/y = ratio Ratios can be applied to: Two different things (ex -- miles and gallons); Two of the same thing (ex -- patients who received a therapy and those who did not)

18 Proportions Type of ratio Numerator is always a subset of the whole Denominator is always whole set Always express relationship between two counts of the same thing Formula: x/(x+y) = proportion

19 Proportion ex: 90/(90+10) = 90/100 = 0.90 Figure 11-8 Proportion of patients given aspirin to a whole set of AMI patients.

20 Rates Measure events occurring over a period of time or to express ratio or proportion as percentage Numerator: 30 AMI patients for which primary PCI done within 90 minutes Denominator: 48 total AMI patients who received a primary PCI Rate ex: 30/48 = = 62.5%

21 Measures of Central Tendency Continuous Variable values being measured, such as time Frequency of distribution how frequently data occurs at given points along continuous variable Central Tendency distribution of a variable measured by: Mean Median Mode

22 Figure Table of AMI patients Time Until PCI sorted by minutes.

23 Mean Average, or sum of values divided by the frequency Can be influenced by extreme values and outliers Example: the sum of the values (1,650 minutes) divided by the frequency (30 patients) 1,650/30 = 55 minutes/patient

24 Median Midpoint in a group of ranked values that divides the data into two equal parts For odd numbers, median is value at center of list For even numbers, median is average of center two values Not influenced by extreme values and outlier cases Example = 59

25 Mode A value that occurs most often in frequency of distribution Not influenced by extreme values and outlier cases Number may not be unique; for example, a sample could contain several values for which there were an equal number of cases Example = 45

26 Figure Timeline showing distribution of AMI cases receiving PCI

27 Measures of Variability Range Variance Standard Deviation

28 Range Difference from highest value in the frequency distribution to the lowest Formula: Maximum Minimum = Range Example: = 73

29 Variance Average variation from the mean Steps: 1. Determine mean; example: Subtract mean from each item in frequency distribution and square the result: (90 55) 2 = 1,225 and so forth 3. Repeat for each case in set, then sum the totals; example: 12, To determine variance, divide sum by number of cases minus 1: 12,812/(30 1) =

30 Standard Deviation Reduces the variance back to the same units as the original values The square root of variance Example: square root of = ~21

31 Pay For Performance (P4P) New development by CMS and other payers Ties reimbursement to improvements in quality Example: CMS s Hospital Quality Initiative links reporting of NHQM to hospital payments for each discharge CMS also developing for physicians and nursing home care Doctors who meet or exceed performance standards receive bonus payments

32 Data Mining in Healthcare Finding previously unknown patterns and trends in healthcare data Motivation for use: Detection of fraud and abuse Processing huge amounts of data from healthcare transactions Generation of useful information for multiple parties, such as hospitals, physicians, and patients And so forth

33 Data Mining Methodology 1. Business understanding Identify business objective and success criteria for data mining project 2. Data understanding and preparation Sampling and data transformation 3. Modeling Data analysis 4. Evaluation Comparison of models and results 5. Deployment Implementation and operationalization of the data mining models

34 Data Mining Techniques Description and Visualization Detecting hidden patterns in data Represent data understanding in the methodology Association and Clustering Association: determine which variables go together Clustering: group objects (patients) so similar patients are grouped together Classification and Estimation Classification: prediction of a categorical target variable (ex predicting healthcare fraud vs. non-fraud) Estimation: prediction of a metric (interval or ratio) target variable (ex length of stay or amount of resources allocation)

35 Data Mining Applications Treatment Effectiveness Compare and contrast causes, symptoms, and courses of treatments Determine which courses of action prove effective Help to standardize treatments of specific diseases

36 Data Mining Applications (Cont.) Healthcare Management Determine ways to better identify and track chronic diseases and high-risk patients Devise appropriate interventions Reduce the number of hospital admissions and claims

37 Data Mining Applications (Cont.) Customer Relationship Management Determine patient preferences, usage patterns, and current/future needs Predict whether a patient is likely to comply with prescribed treatments Fraud and Abuse Establish norms Identify unusual or abnormal patterns of claims Highlight inappropriate prescriptions or referrals and fraudulent insurance and medical claims

Merit-Based Incentive Payment System (MIPS): ST-Elevation Myocardial Infarction (STEMI) with Percutaneous Coronary Intervention (PCI) Measure

Merit-Based Incentive Payment System (MIPS): ST-Elevation Myocardial Infarction (STEMI) with Percutaneous Coronary Intervention (PCI) Measure Measure Information Form 2019 Performance Period 1 Table of