medicaid_utils.adapted_algorithms.py_pmca package

Submodules

medicaid_utils.adapted_algorithms.py_pmca.pmca module

This program implements the Pediatric Medical Complexity Algorithm (Tamara D. Simon, Dorothy Lyons, Peter Woodcox et al 2014) to identify children with complex and non-complex chronic conditions using Medicaid claims data and to distinguish them from children with neither chronic nor chronic complex conditions (healthy children). ‘Complex’ and ‘Non-Complex’ designations are assigned based on whether a child’s condition(s) identified by ICD-9 code(s) can be considered chronic, malignant, or progressive, and whether multiple body systems are involved.

class medicaid_utils.adapted_algorithms.py_pmca.pmca.PediatricMedicalComplexity

Bases: object

This class packages functions to create indicator variables for conditions of interest based on their chronic nature identified by Seattle Children’s Research Institute subset from conditions defined by diagnostic codes outlined in the Chronic Illness and Disability Payment System (CDPS) version 5.3

classmethod create_pmca_condition_counts(df: DataFrame, diag_cd_lst_col: str) → DataFrame

Create PMCA condition counts from diagnosis codes.

For each PMCA body system condition and progressive conditions, counts how many diagnosis codes in the list match, and creates binary indicator columns (any_{condition}, {condition}_2h).

Parameters:

• df (dask.DataFrame) – Patient-level DataFrame with a diagnosis code list column.

• diag_cd_lst_col (str) – Name of the column containing comma-separated diagnosis codes.

Returns:

DataFrame with condition count and indicator columns appended.

Return type:

dask.DataFrame

Examples

>>> # Requires a dask DataFrame with diagnosis code list column
>>> df = PediatricMedicalComplexity.create_pmca_condition_counts(
...     df, 'LST_DIAG_CD_RAW')

data_folder = '/home/runner/work/medicaid-utils/medicaid-utils/medicaid_utils/adapted_algorithms/py_pmca/data'

filename = 'pmca.py'

classmethod get_pmca_chronic_condition_categories(df: DataFrame) → DataFrame

Assign PMCA chronic condition categories.

Creates cond_less (less conservative) and cond_more (more conservative) category columns with values 1 (non-chronic), 2 (non-complex chronic), or 3 (complex chronic).

Parameters:

df (dask.DataFrame) – DataFrame with PMCA condition indicator columns from create_pmca_condition_counts.

Returns:

DataFrame with cond_less and cond_more columns appended.

Return type:

dask.DataFrame

Examples

>>> # Requires a dask DataFrame with PMCA condition indicator columns
>>> df = PediatricMedicalComplexity.get_pmca_chronic_condition_categories(
...     df)

package_folder = '/home/runner/work/medicaid-utils/medicaid-utils/medicaid_utils/adapted_algorithms/py_pmca'

medicaid_utils.adapted_algorithms.py_pmca.pmca.pmca_chronic_conditions(df: DataFrame, diag_cd_lst_col: str = 'LST_DIAG_CD_RAW') → DataFrame

This function implements the Pediatric Medical Complexity Algorithm to identify children with complex and non-complex chronic conditions using Medicaid claims data and to distinguish them from children with neither chronic nor chronic complex conditions (healthy children).

‘Complex’ and ‘Non-Complex’ designations are assigned based on whether a child’s condition(s) identified by ICD-9 code(s) can be considered chronic, malignant, or progressive, and whether multiple body systems are involved.

Definitions of the categories assigned by the Algorithm:

• The less conservative version (cond_less) calculates values as

  • ‘Complex Chronic’:

    1. more than one body system is involved, or

    2. one or more conditions are progressive, or

    3. one or more conditions are malignant

  • ‘Non-complex Chronic’:

    1. only one body system is involved, and

    2. the condition is not progressive or malignant

  • ‘Non-Chronic’:

    1. no body system indicators are present, and

    2. the condition is not progressive or malignant

• The more conservative version (cond_more) calculates values as

  • ‘Complex Chronic’:

    1. more than one body system is involved, and each must be indicated in more than one claim, or

    2. one or more conditions are progressive, or

    3. one or more conditions are malignant

  • ‘Non-complex Chronic’:

    1. only one body system is indicated in more than one claim, and

    2. the condition is not progressive or malignant

  • ‘Non-Chronic’:

    1. no body system indicators are present in more than one claim, and

    2. the condition is not progressive or malignant

Body Systems of interest and the variables used to indicate them:

Body system	variable
cardiac	cardiac
craniofacial	cranio
dermatological	derm
endocrinological	endo
gastrointestinal	gastro
genetic	genetic
genitourinary	genito
hematological	hemato
immunological	immuno
malignancy	malign
mental health	mh
metabolic	metab
musculoskeletal	musculo
neurological	neuro
pulmonary-respiratory	pulresp
renal	renal
ophthalmological	opthal
otologic	otol
otolaryngological	otolar

Parameters:

• df (dask.DataFrame) – Patient level dask dataframe

• diag_cd_lst_col (str, default=LST_DIAG_CD_RAW) – Column name that has comma separated list of diagnosis codes in the observation period

Return type:

dask.DataFrame

Examples

>>> # Requires a patient-level dask DataFrame with diagnosis code list
>>> df = pmca_chronic_conditions(df, 'LST_DIAG_CD_RAW')