Frequent Users of US Emergency Departments
We performed a secondary analysis on publicly available data from the US National Health Interview Survey (NHIS), conducted annually by the National Center for Health Statistics. The NHIS is an annual cross sectional household survey that approximates the non-institutionalised US civilian population. We received a waiver from our institutional review board to analyse data from the 2004–2009 NHIS.
The NHIS collected household interview data from 2004 to 2009 from a total of 157 818 adults (age ≥18 years) who represent an annualised US population of 219 million. The specific details of the NHIS have been described elsewhere. In brief, the sample is obtained by using a stratified multi-stage probability study design with unequal probabilities of selection. Specific subgroups of peoples are purposefully oversampled by the NHIS, including racial/ethnic minorities. New households were surveyed each year, with each year's cohort selected to estimate health and healthcare characteristics of the entire US population. The annual response rate of NHIS is approximately 90% of the eligible households in the sample. Strategies for sampling and methodologies for data collection were very similar to maintain consistency and facilitate comparisons throughout the selected NHIS years.
The NHIS queried the number of ED visits per year by the question, "During the past 12 months, how many times have you gone to a hospital emergency room (this includes emergency room visits that resulted in a hospital admission)?" (none, 1, 2–3, 4–5, 6–7, 8–9, 10–12, 13–15, 16 or more, don't know, refused). There is no commonly agreed upon definition of frequent ED use, with previous literature defining frequent utilisation as three visits annually to 12 or more visits annually. Based on available thresholds of the NHIS categorical data, we chose common definitions of frequent ED utilisation to define our study groups as non-users (0 visits/year), infrequent users (1–3 visits/year), frequent users (4–9 visits/year) and super-frequent users (≥10 visits/year).
We compared demographic, socioeconomic status, health conditions and access to care among these ED utilisation groups. Demographic data included age, sex, marital status and race/ethnicity. Socioeconomic data included employment status, poverty–income ratio and education. Poverty income ratio was defined as per the NHIS as the ratio of family income to the poverty threshold for a family of that size. Self-reported health status was measured by response to the question, "Would you say your health in general is excellent, very good, good, fair or poor?" Cigarette and alcohol use were ascertained, but NHIS did not reliably measure other types of substance abuse. The survey asked about specific chronic health conditions (listed in Table 2) based on the relatively high prevalence in the US adult population and potential for increased primary care and ED utilisation.
To ascertain what type of healthcare facility respondents most often visit for illnesses, participants were asked, "Is there a place that you usually go to when you are sick or need advice about your health?" and "What kind of place do you go to most often—a clinic, doctor's office, emergency room, hospital outpatient department or some other place?" For health insurance, we considered adults with private insurance (with or without any other types of health insurance) as 'private', and persons with Medicaid with or without Medicare as 'Medicaid ± Medicare'. 'Medicare only' consisted only of those with Medicare but without private insurance or Medicaid. 'Other' insurance included adults who do not have private insurance, Medicaid or Medicare, but had other public or military health insurance. The number of outpatient and mental healthcare visits during the past 12 months were also extracted to compare ED utilisation with outpatient and mental health resource utilisation. Missing data were separately coded if prevalence of missingness was ≥2% (eg, for poverty–income ratio) and was dropped if <2% of data were missing.
We performed statistical analyses using Stata V.10.1 (College Station, Texas, USA). Survey commands were used to adjust for the complex survey design and assign population sampling weights. The primary analysis was descriptive with 95% CIs, comparing demographic and clinical characteristics by frequency of ED utilisation. We recognise that many selected variables are inter-related. We used logistic regression to evaluate the association between patient sociodemographic, clinical characteristics and number of ED visits. Two models were constructed, one with the outcome of interest being ≥4 ED visits and the other with the outcome of interest being ≥10 ED visits.
Methods
Study Design
We performed a secondary analysis on publicly available data from the US National Health Interview Survey (NHIS), conducted annually by the National Center for Health Statistics. The NHIS is an annual cross sectional household survey that approximates the non-institutionalised US civilian population. We received a waiver from our institutional review board to analyse data from the 2004–2009 NHIS.
Study Setting and Population
The NHIS collected household interview data from 2004 to 2009 from a total of 157 818 adults (age ≥18 years) who represent an annualised US population of 219 million. The specific details of the NHIS have been described elsewhere. In brief, the sample is obtained by using a stratified multi-stage probability study design with unequal probabilities of selection. Specific subgroups of peoples are purposefully oversampled by the NHIS, including racial/ethnic minorities. New households were surveyed each year, with each year's cohort selected to estimate health and healthcare characteristics of the entire US population. The annual response rate of NHIS is approximately 90% of the eligible households in the sample. Strategies for sampling and methodologies for data collection were very similar to maintain consistency and facilitate comparisons throughout the selected NHIS years.
Data Collection and Measurements
The NHIS queried the number of ED visits per year by the question, "During the past 12 months, how many times have you gone to a hospital emergency room (this includes emergency room visits that resulted in a hospital admission)?" (none, 1, 2–3, 4–5, 6–7, 8–9, 10–12, 13–15, 16 or more, don't know, refused). There is no commonly agreed upon definition of frequent ED use, with previous literature defining frequent utilisation as three visits annually to 12 or more visits annually. Based on available thresholds of the NHIS categorical data, we chose common definitions of frequent ED utilisation to define our study groups as non-users (0 visits/year), infrequent users (1–3 visits/year), frequent users (4–9 visits/year) and super-frequent users (≥10 visits/year).
We compared demographic, socioeconomic status, health conditions and access to care among these ED utilisation groups. Demographic data included age, sex, marital status and race/ethnicity. Socioeconomic data included employment status, poverty–income ratio and education. Poverty income ratio was defined as per the NHIS as the ratio of family income to the poverty threshold for a family of that size. Self-reported health status was measured by response to the question, "Would you say your health in general is excellent, very good, good, fair or poor?" Cigarette and alcohol use were ascertained, but NHIS did not reliably measure other types of substance abuse. The survey asked about specific chronic health conditions (listed in Table 2) based on the relatively high prevalence in the US adult population and potential for increased primary care and ED utilisation.
To ascertain what type of healthcare facility respondents most often visit for illnesses, participants were asked, "Is there a place that you usually go to when you are sick or need advice about your health?" and "What kind of place do you go to most often—a clinic, doctor's office, emergency room, hospital outpatient department or some other place?" For health insurance, we considered adults with private insurance (with or without any other types of health insurance) as 'private', and persons with Medicaid with or without Medicare as 'Medicaid ± Medicare'. 'Medicare only' consisted only of those with Medicare but without private insurance or Medicaid. 'Other' insurance included adults who do not have private insurance, Medicaid or Medicare, but had other public or military health insurance. The number of outpatient and mental healthcare visits during the past 12 months were also extracted to compare ED utilisation with outpatient and mental health resource utilisation. Missing data were separately coded if prevalence of missingness was ≥2% (eg, for poverty–income ratio) and was dropped if <2% of data were missing.
Data Analysis
We performed statistical analyses using Stata V.10.1 (College Station, Texas, USA). Survey commands were used to adjust for the complex survey design and assign population sampling weights. The primary analysis was descriptive with 95% CIs, comparing demographic and clinical characteristics by frequency of ED utilisation. We recognise that many selected variables are inter-related. We used logistic regression to evaluate the association between patient sociodemographic, clinical characteristics and number of ED visits. Two models were constructed, one with the outcome of interest being ≥4 ED visits and the other with the outcome of interest being ≥10 ED visits.
SHARE
