Statistical Methods and Measurement Caveats Explained
When examining statistics related to serious mental illness (SMI) diagnosis, it is crucial to understand the sources of this data and their inherent limitations. Two primary sources for national estimates on mental health in the United States are the National Survey on Drug Use and Health (NSDUH) and the National Comorbidity Survey Adolescent Supplement (NCS-A). This article will delve into the methodologies of these surveys, highlighting their strengths and weaknesses in providing insights into Serious Mental Illness Diagnosis.
National Survey on Drug Use and Health (NSDUH)
Diagnostic Assessment: How SMI is Evaluated
The NSDUH estimates for both Any Mental Illness (AMI) and Serious Mental Illness (SMI) are derived from a predictive model. This model was built using data from clinical interviews conducted with a subset of NSDUH participants (nearly 5,000 adults between 2008 and 2012). These interviews employed a modified version of the Structured Clinical Interview for DSM-IV-TR Axis I Disorders (SCID-I/NP), adapted to assess mental health conditions within the past 12 months. The severity of functional impairment, a key factor in differentiating between AMI and SMI, was evaluated using the Global Assessment of Functioning Scale (GAF).
The diagnostic modules within the SCID-I/NP covered a range of mental health conditions, including:
- Mood disorders: Major depressive episode, manic episode, and others.
- Anxiety disorders: Panic disorder, generalized anxiety disorder, posttraumatic stress disorder, and more.
- Eating disorders: Anorexia nervosa and related conditions.
- Impulse control disorders: Intermittent explosive disorder, adjustment disorder, among others.
- Psychotic symptom screen: To identify potential psychotic features.
It’s important to note that the NSDUH assessment had limitations. It did not include diagnostic modules for conditions such as adult Attention Deficit Hyperactivity Disorder (ADHD), autism spectrum disorders, schizophrenia, or other psychotic disorders (beyond the symptom screen). Consequently, individuals whose only mental health conditions fall into these uncategorized areas might not be fully represented in NSDUH’s AMI and SMI estimates. However, the known high rates of comorbidity among mental disorders suggest that individuals meeting criteria for AMI and/or SMI are likely to be captured, as they may also present with co-occurring disorders that were assessed by the SCID-I/NP.
Population Represented in NSDUH Data
The NSDUH aims to represent the civilian, non-institutionalized adult population (aged 18 and older) residing in the United States. This encompasses individuals living in various housing situations, including:
- Traditional households (houses, apartments, condos).
- Civilian housing on military bases.
- Non-institutional group quarters (shelters, rooming houses, college dorms, migratory worker camps, halfway houses).
However, certain populations are excluded from the NSDUH survey, which is vital to consider when interpreting data related to serious mental illness diagnosis:
- Individuals with no fixed address throughout the year (homeless or transient individuals not in shelters).
- Active-duty military personnel.
- Residents of institutional group quarters (correctional facilities, long-term hospitals).
These excluded groups may have individuals experiencing AMI and/or SMI, who are not accounted for in the NSDUH estimates. Furthermore, data on respondent sex was categorized solely into male and female.
Survey Non-response and Potential Biases
In 2022, a significant 52.0% of the selected NSDUH sample of adults did not complete the interview. This non-response rate has increased compared to pre-2020 levels, potentially impacting the reliability of comparisons. Reasons for non-response include:
- Refusal to participate (29.3%).
- Respondent unavailability (18.2%).
- Other factors like physical/mental incapacity or language barriers (4.5%).
It is plausible that individuals with mental illnesses, including serious mental illnesses, may be disproportionately represented in these non-response categories. While NSDUH weighting adjustments are implemented to mitigate bias from non-response, they might not fully compensate for differential non-response based on mental health status, potentially affecting the accuracy of serious mental illness diagnosis prevalence estimates.
Data Suppression for Reliability
Data for certain demographic groups may be suppressed in NSDUH reports due to concerns about low precision. Suppression occurs when data do not meet acceptable thresholds for prevalence estimates, standard error estimates, and sample size, ensuring only statistically sound data is presented regarding serious mental illness diagnosis.
Impact of the COVID-19 Pandemic on 2022 NSDUH Data
The COVID-19 pandemic brought changes to NSDUH data collection in 2022. Multimode data collection (both in-person and virtual) was continued from the fourth quarter of 2020. In 2022, 40.7% of interviews were web-based, and 59.3% were in-person. Response rates in 2022 were notably lower, with an overall response rate of 12.1% for people aged 12 or older. These methodological changes and lower response rates necessitate caution when comparing 2022 NSDUH estimates with previous years, especially when analyzing trends in serious mental illness diagnosis.
For in-depth information on data collection and calculation methods, refer to the 2022 National Survey on Drug Use and Health Methodological Summary and Definitions report.
National Comorbidity Survey Adolescent Supplement (NCS-A)
Diagnostic Approach and Target Group: Youth Mental Health
The NCS-A, sponsored by the National Institute of Mental Health (NIMH), was designed to provide national data on the prevalence and correlates of mental disorders among youth in the U.S., a crucial demographic for understanding the early onset of serious mental illness diagnosis. This face-to-face survey included 10,123 adolescents aged 13 to 18 in the continental United States, conducted between February 2001 and January 2004.
The NCS-A employed a dual-frame design:
- Adolescents from households participating in the adult U.S. National Comorbidity Survey Replication (904 participants).
- Adolescent students from a nationally representative sample of 320 schools (9,244 participants).
DSM-IV mental disorders were assessed using a modified version of the World Health Organization Composite International Diagnostic Interview, a fully structured instrument ensuring standardized diagnostic criteria for mental health conditions relevant to serious mental illness diagnosis in youth.
Survey Participation Rates and Potential Bias in Adolescent Data
The overall adolescent non-response rate for the NCS-A was 24.4%. This varied across sampling frames:
- Household sample: 14.1% non-response.
- Un-blinded school sample: 18.2% non-response.
- Blinded school sample: 77.7% non-response (primarily due to parental consent not being returned).
Refusal to participate was the main reason for non-response (21.3%), largely originating from parents in household and un-blinded school samples (72.3% and 81.0% respectively). In the blinded school sample, refusals were almost entirely due to parents not returning consent postcards (98.1%). These non-response patterns could introduce biases, potentially affecting prevalence estimates for serious mental illness diagnosis among adolescents.
For further details, see PMID: 19507169 and the NIMH NCS-A study page.
Last Updated: September 2024
