Concept: Alternate Level of Care (ALC) Patients - Method of Identification

Introduction

An Alternate Level of Care (ALC) patient is someone who is occupying an acute care hospital bed but not acutely ill or does not require the intensity of resources or services provided in a hospital setting (e.g.: may no longer need treatment or hospital care). ALC patients are primarily those who no longer need acute care but still require some form of care such as personal care homes (PCH) placement, home care or other forms of service. Alternate Level of Care (ALC) is a concern for two major reasons:


• For patients, it means that they are not in the most appropriate place for the type of care they need.
  
• For hospitals, it means that needed beds may not be available, causing back-ups in the system such as longer waits for elective procedures or admissions from the emergency department. (Who is in Our Hospitals ... and Why? by Fransoo et al., 2013).

This concept contains the following information, based on research completed for the deliverable by Fransoo et al. (2013):


• a description of the two methods used to identify ALC patients in hospital;
  
• an algorithm that describes the steps in identifying ALC patients;
  
• a Results/Findings section which provides links to important tables and figures related to ALC that are available in the deliverable; and
  
• a section that describes some cautions / limitations when working with ALC data.

Methods

The Hospital Abstracts Data collects information from every inpatient stay in Manitoba hospitals. In Fransoo et al. (2013), they categorized patients into the following main service types: Alternate Level of Care (ALC), Medical, Surgical, Pregnancy and Birth, Mental Diseases and Disorders, and Palliative Care. In this research, patients identified as ALC were a particular focus because a significant portion of hospital days are dedicated to providing this type of care to patients.

Beginning in 2004/05, the new Hospital Abstract in Manitoba hospitals allowed for the routine collection of ALC related data. However, the initial collection of the data was not consistent throughout all hospitals, and the coding of ALC patients may have been under reported. As a result, in Fransoo et al. (2013) they developed two methods to identify ALC patients: the first using the abstract coding rules for ALC patients, and a second that identified possible ALC patients based on other criteria. Both methods are described below.

1. Coded ALC Patients

All ALC patients should have the following coded on their hospital abstract:


• at least one ALC Reason Code;
  
• at least one ICD-10-CA diagnosis Z code that indicates the reason for ALC designation; and
  
• a Main Patient Service Code or a Service Transfer Code = 99 (ALC).

Additional criteria include:


• ALC patients must be designated ALC by their attending physician or an authorized delegate (Canadian Institute for Health Information (CIHI), 2010).
  
• an acute level of care (LOC) code (0-6) should be recorded. As of April 1, 2010, some non-acute levels of care are permitted to code ALC. These include: Chronic (code 3), Mental Health (code 5) and Rehab (code 2).

ALC Reason Codes

In Manitoba there are five major categories of ALC Reason Codes for why a patient is designated as ALC and each patient can be assigned up to three ALC Reason Codes in one hospitalization. The group code values and description for each of the major categories are:


• 10 - Panel Process for possible placement in PCH or Chronic Care facility;
  
• 20 - Waiting for Home Care or Community Care Services;
  
• 30 - Waiting for other placement such as a group home, supportive housing or hospice;
  
• 40 - Waiting due to other reasons such as respite, as a boarder baby or boarder mother or awaiting support or accommodation; and
  
• 50 - Rehabilitation Services (WHRA only).

Each major category has specific detailed codes that should be recorded in the abstract. Major group code values should not be recorded in the abstract. For a complete listing of all ALC Reason Codes see the ALC Reason Code definition in Fransoo et al. (2013).

On/Off Dates and ALC Length of Stay (ALCLOS)

For each ALC Reason Code there should be a corresponding set of "on and off" (start and stop) dates. These are recorded in the variables: alc_ondt1 to alc_ondt3 and alc_offdt1 - alc_offdt3. These variables can be used to calculate the ALC length of stay (ALCLOS). Beginning with 2007/2008 hospital data, a variable ALCLOS is also available. ALCLOS is the total number of hospital days designated as ALC for that hospital stay.

ALC-Related Z Codes

The Z codes are part of the International Classification of Diseases (ICD-10-CA) diagnoses codes that fall under the category "Factors influencing health status and contact with health services." These codes do not assign a diagnosis due to illness or injury, but rather provide supplementary information as to the patient's status and/or reason for admittance to or remaining in the hospital. Examples include routine physical examinations, vaccinations or outcome of delivery.

There are certain Z codes that are related to Alternate Level of Care (ALC), but they are not limited to ALC designation and may be used for any patient. Every patient designated ALC should have an ALC-related Z code in their hospital abstract to indicate the reason the patient was designated ALC. Coded ALC patients may have one or more ALC-related Z code recorded to indicate the reason for ALC designation (Canadian Institute of Health Information, 2010).

For a listing of ALC-related Z codes used in Fransoo et al. (2013), see the Z Code definition.

2. Possible ALC Hospitalizations

"Hospitalizations or days of hospital care categorized as Possible ALC (rather than Coded ALC) are those for which the patient did not have an ALC Reason Code included on their hospital abstract, and was not designated as Alternate Level of Care (ALC), but due to markers in the data may possibly have been a non-acute patient for some portion of their acute care hospital stay. Due to possible under-coding of ALC patients, it may be the case that properly coded ALC hospitalizations and days of care are a subset of the true number of non-acute patients occupying acute care hospital beds." (Fransoo et al, 2013).

In an attempt to identify additional hospitalizations and days in care that were likely non-acute, or possibly ALC, several algorithms were employed, in a hierarchical fashion, to identify these patients. Additional data files were used to assist in this investigation, including the Long Term Care (LTC) Utilization History Data, Home Care Utilization History Data and Home Care Minimum Data Sets (MDS) Assessment Data. The Need to Know Team project Team identified 4 different possible scenarios for non-coded ALC patients:


1. PCH Admissions - these are patients directly discharged from hospital into a PCH and not coded as ALC in their abstract;
   
2. New Home Care Cases - this includes people who did not have an open Home Care case on the day they were admitted to hospital but did on the day they were discharged from hospital. People in this group most likely had their discharge delayed somewhat, due to the process of arranging Home Care. This may not be true in all cases, but some of these patients should have been coded ALC during their stay.
   
3. Diagnoses codes - some Z codes are closely related to ALC patients, and patients were labelled Possible ALC when they had not been coded as ALC, but did receive an ALC-related Z code. NOTE: Table 3.4 shows the distribution of ALC-related Z codes used to identify possible ALC cases.
   
4. Diagnoses Codes and Service Transfer Codes - some of the patients in group 3 Diagnosis codes listed above also had Service Transfer Codes that specified the number of days of hospital care spent in each Service Type. These codes and dates determined the number of Possible ALC days of care involved.

For more detailed information on identifying possible ALC hospitalizations, please read the section titled Possible ALC Hospitalizations and Days of Care in Fransoo et al. (2013).

ALC Algorithm Hierarchy

The algorithm used to identify all ALC patients and ALC days of care, including both coded ALC cases and possible ALC cases is:


1. Identify ALC patients using the Coded ALC method described above, looking at ALC reason codes, ALC start date and Z codes.
   
2. Patients assessed for PCH placement during their hospital stay, and then admitted to PCH within ± 3days of discharge will have the time in hospital from panel date onward allocated as ALC days. Additional days will be allocated for time waiting to be paneled. This was allocated as 24% of patient's stay based on fully-coded ALC patients with ALC reasons 11 and 12 coded).
   
3. Patients with a new home care case opened within ± 5days of discharge will have a portion of the time in hospital allocated as ALC days.
   
4. Patients with an ALC-related Z code and corresponding diagnosis type W, X or Y (for 1st, 2nd or 3rd service transfer diagnosis) will have the service transfer days allocated as ALC days.
   
5. Patients with only an ALC-related Z-code will have a portion of their hospital stay allocated as ALC days. The number of days allocated is specific to each Z-code, and is based on the proportion of time that fully-coded ALC patients with the same Z-code spent in hospital as ALC rather than acute. Column 3 (Percent of Hospitals Days Allocated as ALC) in Table 3.4 lists the percent of total days allocated to ALC for these cases.

Results and Findings

The following tables and figures are presented in Fransoo et al. (2013):


• Appendix 4 contains a number of tables that provide details on the average annual number of hospitalizations and days of hospital care by patient's region of residence (RHA) by main service type (including coded ALC and Possible ALC categories), and by different category of hospitals (all Manitoba hospitals, regional hospitals, type of hospital and by individual hospital).
  
  
• Figure 3.7 & 3.8: ALC Hospitalizations and Days of ALC Hospitalization by Region of Hospital location
  
  
• Figure 3.9 & 3.10: Where RHA Residents Went for ALC Hospitalizations and Days of ALC Care
  
  
• Table 3.3: Logistic Regression: Probability of Fully Coded ALC Hospitalization - this table lists the covariates, adjusted odds ratio and p-values for the factors used in the regression analysis.
  
  
• Table 3.4: Distribution of ALC-Related Z Codes Used to Identify Possible ALC Cases Among Acute Care Hospitalizations - this table provides an overview of the Z codes and the Manitoba ALC Reason Codes. The percent found in the 3rd column were used to calculate Possible ALC days of care among patients who are coded with Z codes but are not coded as ALC.
  
  
• Table 3.5: Possible ALC Hospitalizations and Days of Hospital Care Identified by Data Source - lists the total number of hospitalizations and days of care calculated for each scenario in identifying possible ALC cases, and overall total numbers.
  
  
• Table 4.1: Hospitalizations Summary by Hospital
  
  
• Table 4.2: Days of Hospital Care Summary by Hospital

Limitations / Cautions

• This study revealed that not all ALC patients may be properly coded as ALC patients in the Hospital Abstracts data. Therefore two methods are used to identify ALC cases: coded and possible ALC patients.