Capacity to Treat Methodology

PurpleLab is presenting risk adjustment cohort related population counts as indicators of likely “demand” (numerators) while comparing treatment capacity counts as indicators of likely “supply” (denominators). PurpleLab considered Capacity-to-Treat as three separate measures:

The method is based on the first three of the following key components:

Capacity of Bed Space:

The number of hospital beds and intensive care unit (ICU) beds

Capacity of Staffing:

The number of Physicians Intensivists (with the necessary experience in managing ventilator dependent patients) Ajao et al. [1] and Gershengorn et al. [2].

Capacity of Supplies:

The number of ventilators and ancillary supplies

The Capacity of Supplies component of the model is assumed implied in the Capacity of Bed Space component. Specifically, in the number of ICU beds. That is, the final quantification of healthcare capacity is based on both the Capacity of Bed Space and the Capacity of Staff components of the model.

With respect to the Capacity of Bed Space component of the model (total hospital beds and ICU beds), we obtained the respective data from the American Hospital Association’s for-profit subsidiary Health Forum.

PurpleLab coded the AHA Hospitals to our proprietary healthcare organization provider taxonomy and then selected those hospital types represented in Table 1 for inclusion and those hospital types, represented in Table 2 for exclusion. This reduced the total AHA “hospital” inventory to 4,953 hospitals and eliminated “hospitals” that would be inadequately prepared to care for ventilator dependent patients.

Table 1. Hospital types included in the Capacity of Bed Space counts

Table 2. Hospital types excluded in the Capacity of Bed Space counts

With respect to the Capacity of Staff component of the model (specialized medical personnel), we obtained the respective data from the National Plan & Provider Enumeration System (NPPES).

The NPPES issues the National Provider Identifier (NPI), a unique 10-digit number identifier issued to healthcare providers by the Health and Humans Services, Centers for Medicare and Medicaid Systems (CMS) Agency. When providers apply for an NPI, they must select a taxonomy code indicating a provider type and specialty type. They can select multiple taxonomy codes.

Taxonomy codes used in counting of Physician Intensivists are presented in Table 3.

Table 3. Physician Intensivists taxonomy codes used in the Capacity of Staffing counts:

With respect to mapping Capacity-to-Treat counts to the proper 3-digit Zip code region (ZIP3), PurpleLab relied on its maintenance of “location” information for all healthcare providers, including Hospitals and Physician Intensivists. We define “locations” as an address location.

PurpleLab’s Provider Locations API connects provider entities to up-to-date location information. PurpleLab has developed an end-to-end process that integrates more than 350 separate provider databases each month to include address standardization (according to USPS CASS Certification standards), “street level” geocoding (leveraging best in class Google Map’s API) and meta-data tracking that enables PurpleLab to map providers to their “best” locations.

We consolidate duplicated locations, flag “residential” locations, flag “mailing only” locations and implement business rules to select the likely “best” primary practice locations as well as all valid secondary practice locations. We used the “best” primary practice location information available to us to map the Hospital and Physician Intensivist providers to the appropriate ZIP3 regions.

References:

1.  Ajao A, Nystrom SV, Koonin LM, et al. Assessing the Capacity of the US Health Care System to Use Additional Mechanical Ventilators During a Large-Scale Public Health Emergency. Disaster Med Public Health Prep. 2015;9(6):634–641. doi:10.1017/dmp.2015.105.

2.  Hayley B. Gershengorn, MD; David A. Harrison, PhD; Allan Garland, MD, MA; et alAssociation of Intensive Care Unit Patient-to-Intensivist Ratios With Hospital Mortality. JAMA Intern Med. 2017;177(3):388-396. doi:10.1001/jamainternmed.2016.8457