Concept: Undercounting Hospital Visits

Introduction

Ambulatory visits definition has been simplified and can be generated using a SAS macro on MCHP systems. Most of this concept is retained for historical purposes. See the Ambulatory Visits - Physicians concept for more recent information.

Based on a March 27, 2002 researchers meeting, the following definition for hospital visit was accepted.


• Service Date falls between or on Admission and Separation date irrespective of what hospital is coded.
  
• If there is an OPD/ER flag set to 'O','E' and the physician service is on the date of admission or separation it is counted as an ambulatory visit not in-hospital.
  
  
• See SAS code below (internal access only) for linking with hospital data and defining ambulatory visits.

The current ambulatory visits (ambvis='1') misses hospital visits for outpatient cases. These visits are misclassified as in-patient visits.


• In-patient visits are defined as a contact with a physician that occurs during a time period where a patient has been admitted to a hospital.
  
• Hence a true in-patient contact should have a corresponding hospital discharge abstract.

In 1992, it was noted that "we may miss visits that did not occur when the patient was admitted to hospital". It was estimated that "this may represent a 2% under-estimation of ambulatory claims overall".

• We linked the in-patient visit definition from the medical services/physician claims data against the hospital discharge abstracts data. Those occurring during a hospital stay are true in-patients.
  
• Those without a corresponding hospital discharge abstract are visits that we have neglected in all our analysis.
  
• Throughout the years we have under-estimated the number of visits.

Hospital visits and magnitude of the under-estimation

• Overall visits: The proportion of the missing hospital visits is growing over-time. While it was 2% in 1992 it reached 4 + % in 1998.
  
• Specialist Visits: The bulk of the missing visits correspond to outpatient visits to specialist. In 1998 the under-estimate reached 14%.
  
• Consults visits: For the period 1993-1998 the under-estimate for consults was between 16 to 23%.
  
• GP visits: less than 1% under-estimate throughout the period.

Income Quintiles

Special attention should be paid to the effect of the under-estimate when dealing with income quintiles.

The lower the income, the higher the proportion of missing visits:

Income Quintile	% Missing Specialist Visits	% Missing Specialist Visits in dollars	% Missing Consult Visits
Q1 (poorer)	16.7	20.5	24.1
Q2	15.3	18.2	20.4
Q3	14.6	17.4	20.4
Q4	13.7	16.1	19.6
Q5 (wealthier)	13.5	15.3	19.7

Implications

• Analysis such as utilization rates or diagnosis analysis need to incorporate these hospitals visits.
• Data sets are available for the period 1993-1998 (hospvis93 to hospvis98)- contact Bogdan for more information.