Publications
HCCI’s original reports powered by #HCCIdata

ADRD Prevalence in Various Insurance Populations: A Collaboration with The Alzheimer’s Association

Alzheimer's disease and related dementias (ADRD) represent a significant and growing cost to the United States health care system. While the prevalence and cost of ADRD related to Medicare Fee-for-Service beneficiaries is documented in the Center for Medicare and Medicaid Services' Chronic Conditions Warehouse, less is known about the prevalence and cost of ADRD among individuals covered by employer-sponsored insurance or Medicare Advantage plans. On behalf of the Alzheimer's Association, HCCI used its commercial claims dataset to examine the prevalence and costs associated with Alzheimer's Disease and related dementias among these populations during the years 2016 and 2017. A more extensive publication exploring the findings of this analysis is forthcoming; the results are provided in downloadable data tables below, along with a summary of our methods. 


Methods Note

Methods Note

Prevalence Data

Sample Selection

Denominator Qualification

For the purpose of this project, HCCI used the base year of 2016 for the prevalence analyses.  HCCI also calculated prevalence for the year of 2017 to be used as a comparison group. 

In order to be included in the denominator for the prevalence, a person must have:

  • At least one month of coverage in the base year
  • Age is not null and greater than or equal to 35 (and less than 65 for the employer-sponsored insurance population)
  • For commercially insured and Medicare Advantage individuals, the insurance product must be either POS, PPO, HMO, or EPO
  • For Medicare Fee-for-Service individuals, the number of Part A months in the year must be equal to the number of Part B months in the year

Numerator Qualification

Once an individual has met the qualifications for the denominator, we use the presence of specific ICD-10 codes to flag an individual as having Alzheimer’s disease related dementia (ADRD). For the purpose of this analysis, we use a subset of ICD-10 codes to define ADRD. Any of the first 10 ICD-10 codes listed on a claim were included. This set was created using the Chronic Conditions Warehouse list and through consultation with Julie Bynum, MD, MPH.  The ICD-10 codes chosen are:

G30.0, G30.1, G30.8, G30.9, G31.01, G31.09, G31.83, F01.50, F01.51, F02.80, F02.81, F03.90, F03.91.

In addition to the 1 year prevalence estimation we estimated the 3-year prevalence to create estimates more comparable to those reported by the CCW (3-year includes base year and any codes present in the previous 2 years, see methodology used in the Chronic Conditions Warehouse).  Since the ICD-10 transition began on October 1, 2015, selected ICD-10 codes were crosswalked to ICD-9 equivalents. Any of the 3 ICD-9 codes on a given claim were included. These ICD-9 codes are:

290.40, 290.41, 290.42, 290.43, 294.10, 294.11, 290.0, 290.10, 290.11, 290.12, 290.13, 290.20, 290.21, 290.3, 290.8, 290.9, 294.20, 294.21, 291.2, 331.0, 331.11, 331.19, 331.82.

Subgrouping

For the cohort identified in the prevalence sample, subgroups are reported based on age, gender, and dual eligibility status (for Medicare Fee-for-Service population).  The following descriptions of the subgroups are below:

Age Groups – For each individual in the base year of interest, the age of the person is determined as the year of their birth subtracted by the base year.  This method is used to ensure individuals do not change age groups within the base year of interest if they have multiple claims related to ADRD.  The variables used in the HCCI commercial claims datasets are “yr” (base year – taken from either the inpatient, outpatient, or physician file containing the ICD-9 or ICD-10 codes of interest) and “ybirth” (year of birth – taken from the member file).  For Medicare Fee-for-Service claims, we used the “bene_age_at_end_ref_yr” which does the calculation of the year of birth subtracted from the base year.  Age groups were divided into the categories of 35-44, 45-54, 55-64, 65-74, 75-84, and 85 and over.

Gender – Gender variable as reported on the patient’s member file.  The variable used in the HCCI commercial claims dataset is “gdr” with the values of ‘1’ for male and ‘2’ for female, with any people containing a null value or ‘9’ within a month having that person-month level data eliminated.  For Medicare Fee-for-Service claims, we used the variable “bene_sex_ident_cd” with values of ‘1’ for male and ‘2’ for female, also eliminating anyone with null or ‘0’ for unknown.

Calculations

1-Year Prevalence

Once datasets were created containing data for all individuals meeting population restriction criteria and variables were attached for subgrouping distinctions, an individual was then flagged as having ADRD within the base year if they contained any of the ICD-10 codes identified in Numerator Qualifications. Once identified we performed the following calculations were performed:

Numerator (Months) –Distinct number of months a person is present in our data if they are flagged for having ADRD within the year for the subgroups of interest.

Denominator (Months) – The number of months for the entire data where people meet the sample restrictions within the year for the subgroups of interest.

Numerator (People) - Number of people present in our data if they are flagged for having ADRD within the year. This variable may also be called “Distinct People with Alzheimer’s” for the subgroups of interest.

Denominator (People) – The number of distinct individuals where people meet the sample restrictions within the year for the subgroups of interest.

Distinct Alzheimer’s Diagnoses – Total number of non-duplicate ICD-10 codes related to ADRD present for an individual within the year for the subgroups of interest.

Prevalence (Months) – ((Numerator(Months)/Denominator(Months)) within a year and subgroups.

Prevalence (People) –  ((Numerator(People)/Denominator(People)) within a year and subgroups.

Population Weights – For Medicare Advantage and Fee-for-Service separately, these population weights are factors to multiply our data results to be able to generalize our sample to all people within those payer categories, broken down by age groups and gender categories.  These weights were only calculated for the base year of 2016.

For individuals with employer-sponsored insurance, American Census Bureau’s 2016 number of individuals with employer-sponsored insurance was divided by the people meeting the population criteria (known gender and year of birth) in our member dataset.  The data from the American Census Bureau can be found here.

For individuals with Medicare Advantage, we took the total number of individuals in our Medicare beneficiary summary file for 2016 with greater than 0 months in an HMO plan (signified by the variable “bene_hmo_cvrage_tot_mons”) divided by the people meeting the population criteria (known gender and year of birth) in our member dataset.

Distinct People with Alzheimer’s Population Weighted – Number of distinct people with ADRD in our dataset multiplied by the population weight to generalize our numbers to the greater employer-sponsored insurance and Medicare Advantage populations.

3-Year Prevalence

All variables in this data sheet are calculated in the same manner, the difference being 3 years of diagnosis codes were included to determine the presence of ADRD diagnoses, as mentioned in the Numerator Qualification.  This change kept the denominators populations the same and increased the numerators since more people were flagged with ADRD with this method.

Dual Status – For only Medicare Fee-for-Service claims, people were identified as dual eligible if they had more than 0 months of dual eligibility, identified using the “bene_state_buyin_tot_mons” variable in the base year.


Cost Cohort Data

Sample Selection

Control Population Qualifications

For the purpose of this project, HCCI used the base year of 2016 for initial population restriction criteria.  Most of the population criteria follow the prevalence cohort criteria, except that individuals with negative claim totals in any year 2015 through 2017 are eliminated from the sample. This usually occurs due to a claim reversal that is given a separate claim first date and last date.  To prevent this reversal from affecting totals, these individuals are fully excluded from the cohort, giving slightly lower population counts than the prevalence cohort.

In order to be included in the denominator for the prevalence, a person must have:

  • At least one month of coverage in the base year of 2016
  • Age is not null and greater than or equal to 35 (and less than 65 for the employer-sponsored insurance population)
  • For commercially insured and Medicare Advantage individuals, the insurance product must be either POS, PPO, HMO, or EPO
  • For Medicare Fee-for-Service individuals, the number of Part A months in the year must be equal to the number of Part B months in 2016.
  • Total spending within a category of spending must be greater than or equal to zero for both 2016 and 2017.

Once an individual meets initial qualifications, they must meet a further set of restrictions in order for their 2017 costs to be calculated.  The number of individuals who are excluded via these criteria are noted by “Lost Patients” and their corresponding months in 2016 are noted by “Lost Patient Months”. The criteria for this step are detailed below:

  • At least one month of medical coverage in 2017.
  • The number of months of medical coverage in 2017 are required to have the same number of months covered by prescription drug coverage (noted by “rx_cvg_ind” variable in commercial claims and “ptd_plan_cvrg_mons” in Fee-for-Service claims)
  • If an individual is flagged as dual eligible or not dual eligible in 2016, the same must be true for 2017.
  • If an individual has a claim for ADRD as noted by ICD-10 codes of interest in 2016, the same must be true for 2017.

After meeting all these criteria, the number of months an individual is present in the 2017 claims are labeled as “Next Year Months”.

ADRD Individual Qualification

Once an individual has met the initial qualifications for the control population, we use the presence of specific ICD-10 codes to flag an individual as having ADRD.  For the purpose of this analysis, we used a subset of select ICD-10 codes to define ADRD.  Any of the first 10 ICD-10 codes listed on a claim were included. This set was created using the Chronic Conditions Warehouse list and through consultation with Julie Byum, MD, MPH.  The ICD-10 codes chosen are:

G30.0, G30.1, G30.8, G30.9, G31.01, G31.09, G31.83, F01.50, F01.51, F02.80, F02.81, F03.90, F03.91.

This population of individuals with ADRD then go through the same further restrictions as described in the Control Population Qualifications with the only caveat being that the individual must have an ADRD claim in both 2016 and 2017.

Subgrouping

The same methods of subgrouping individuals as described in the prevalence methodology are applied to the cost cohort.  The main difference is that the subgrouping is based off of the year of 2016 and carried through to 2017.  This is to associate one age to an individual rather than having the individual change age groups. Also, control and ADRD populations are separated by the adding in a subgroup of an “ADRD Flag” rather than separating the ADRD population as the numerator and the entire population as the denominator, as was done in the prevalence calculations.

Calculations

Cost Calculations

Once datasets were created containing data for all individuals meeting population restriction criteria and variables were attached for subgrouping distinctions (including the ADRD flag distinction), then the following calculations were performed:

Distinct People Counts – Number of distinct individuals in 2016meeting all cost cohort restrictions (both the preliminary and further restrictions) for the subgroups of interest.

Current Year Months – Number of months in 2016 individuals contain after meeting all cost cohort restrictions (both the preliminary and further restrictions) for the subgroups of interest.

Lost Patients – Number of distinct individuals in 2016 meeting the preliminary restrictions but NOT meeting the further restrictions for the subgroups of interest.

Lost Months – Number of months in 2016 for individuals meeting the preliminary restrictions but NOT meeting the further restrictions for the subgroups of interest.

Percent Lost from 2016 to 2017 – (Current Year Months/(Current Year Months + Lost Months)) for the subgroups of interest.

Next Year Months – For the distinct people meeting preliminary and further restrictions (from the distinct people counts variable), the number of months they are present in the 2017 data. In HCCI data, this is taken as the total of the “allwd_amt” variable for the entire year of 2017 for subgroups of interest.

Inpatient allowed amount – For Medicare Advantage and employer-sponsored insurance individuals, the total number of dollars for inpatient care in 2017 that can be associated with people falling in the subgroups of interest and meeting all cost cohort restrictions. In HCCI data, this is taken as the total of the “allwd_amt” variable for the entire year of 2017 for subgroups of interest.

Outpatient allowed amount – For Medicare Advantage and employer-sponsored insurance individuals, the total number of dollars for outpatient care in 2017 that can be associated with people falling in the subgroups of interest and meeting all cost cohort restrictions. In HCCI data, this is taken as the total of the “allwd_amt” variable for the entire year of 2017 for subgroups of interest.

Physician allowed amount – For Medicare Advantage and employer-sponsored insurance individuals, the total number of dollars for professional services care in 2017 that can be associated with people falling in the subgroups of interest and meeting all cost cohort restrictions. In HCCI data, this is taken as the total of the “allwd_amt” variable for the entire year of 2017 for subgroups of interest.

Prescriptions allowed amount – For Medicare Advantage and employer-sponsored insurance individuals, the total number of dollars for prescriptions in 2017 that can be associated with people falling in the subgroups of interest and meeting all cost cohort restrictions. In HCCI data, this is taken as the total of the “allwd_amt” variable for the entire year of 2017 for subgroups of interest.

Outpatient claim payment allowed amount – For Medicare Fee-for-Service individuals, the total number of dollars for outpatient care in 2017 that can be associated with people falling in the subgroups of interest and meeting all cost cohort restrictions.  In FFS data, the sum of “clm_pmt_amt”, “nch_bene_ptb_ddctbl_amt”, and “nch_bene_ptb_coinsrnc_amt” for the entire year of 2017 based on the claim from date for subgroups of interest.

Inpatient claim payment allowed amount – For Medicare Fee-for-Service individuals, the total number of dollars for inpatient care in 2017 that can be associated with people falling in the subgroups of interest and meeting all cost cohort restrictions.  In FFS data, the sum of “clm_pmt_amt” and “nch_ip_tot_ddctn_amt” for the entire year of 2017 based on the claim from date for subgroups of interest.

Carrier claim payment allowed amount – For Medicare Fee-for-Service individuals, the total number of dollars for carrier care in 2017 that can be associated with people falling in the subgroups of interest and meeting all cost cohort restrictions.  In FFS data, the sum of “line_alowd_chrg_amt” for the entire year of 2017 based on the claim from date for subgroups of interest.

Hospice claim payment allowed amount – For Medicare Fee-for-Service individuals, the total number of dollars for hospice care in 2017 that can be associated with people falling in the subgroups of interest and meeting all cost cohort restrictions.  in ffs data, the sum of “clm_tot_chrg_amt” for the entire year of 2017 based on the claim from date for subgroups of interest.

SNF claim payment allowed amount – For Medicare Fee-for-Service individuals, the total number of dollars for skilled nursing facility care in 2017 that can be associated with people falling in the subgroups of interest and meeting all cost cohort restrictions.  In FFS data, the sum of “clm_pmt_amt” and “nch_ip_tot_ddctn_amt” for the entire year of 2017 based on the claim from date for subgroups of interest.

HHA claim payment allowed amount – For Medicare Fee-for-Service individuals, the total number of dollars for home health agency care in 2017 that can be associated with people falling in the subgroups of interest and meeting all cost cohort restrictions.  In FFS data, the sum of “clm_tot_chrg_amt” for the entire year of 2017 based on the claim from date for subgroups of interest.

DME claim payment allowed amount – For Medicare Fee-for-Service individuals, the total number of dollars for durable medical equipment in 2017 that can be associated with people falling in the subgroups of interest and meeting all cost cohort restrictions.  In FFS data, the sum of “line_alowd_chrg_amt” for the entire year of 2017 based on the claim from date for subgroups of interest.

Part D claim payment allowed amount – For Medicare Fee-for-Service individuals, the total number of dollars for Part D claims in 2017 that can be associated with people falling in the subgroups of interest and meeting all cost cohort restrictions.  In FFS data, the sum of “cvrd_d_plan_pd_amt” for the entire year of 2017 based on the claim from date for subgroups of interest.

Total Medical Payment – Total payment for services that can be associated with medical related care (all services not in Part D claim payment for FFS individuals or prescription allowed amount for commercially insured individuals) in 2017 for subgroups of interest.

Total Prescription Drug Payment – Total payment for services that can be associated with prescriptions (Part D claims FFS individuals or prescription allowed amount for commercially insured individuals) in 2017 for subgroups of interest.

Total Payment – Total medical and prescription care payment in 2017 for subgroups of interest.

% Total Payment on Medical – (Total Medical Payment/Total Payment) in 2017 for subgroups of interest.

% Total Payment on Prescriptions – (Total Prescription Drug Payment/Total Payment) in 2017 for subgroups of interest.

Total Payment per Month – (Total Medical Payment/Next Year Months) in 2017 for subgroups of interest.

% Difference from Control Group – (Total Payment per Month for ADRD individuals minus Total Payment per Month for control individuals/Total Payment per Month for control individuals) in 2017 for subgroups of interest.

By accepting you will be accessing a service provided by a third-party external to https://healthcostinstitute.org/