What role does mental health play in preventable hospitalizations? In a new article in the January 2018 issue of Medical Care, Dr. Laura Medford-Davis and colleagues report that in Texas, mental illnesses were associated with higher odds of preventable hospitalizations. Using an administrative database of all Texas hospital admissions from 2005-2008, the authors found that 13% of hospitalizations were potentially preventable.
The Agency for Healthcare Research and Quality (AHRQ) definitions for potentially preventable conditions (also known as ambulatory care sensitive conditions) include a variety of acute and chronic conditions, such as pneumonia, urinary tract infection, congestive heart failure (CHF), and chronic obstructive pulmonary disease (COPD), among others. The new analysis by Medford-Davis and colleagues explores the associations of diagnosed mental illness with preventable hospitalizations, overall and by type of hospitalization. The authors find statistically significant relationships even after controlling for a variety of other factors, such as insurance status, Charlson comorbidity index, age, admission type, and other patient and hospital characteristics.
Why should we worry about the relationship between mental health conditions and preventable hospitalizations? People living with a serious mental illness (SMI) die 13-30 years earlier than the general population on average–a sizable mortality gap. They also have a higher prevalence of some chronic illnesses, such as diabetes, than the non-SMI population. As pointed out by Medford-Davis and colleagues, understanding the role of mental health in potentially preventable hospitalizations might lead to interventions or treatment guidelines to increase the use of appropriate care.
The relationships between mental health and preventable hospitalizations are complex. Certain conditions might lead to more or less use of care. For example, anxiety might lead to preventable hospitalizations by exacerbating COPD symptoms, but could also lead to greater use of preventive care that prevents hospitalizations. Other conditions, such as bipolar disorder, might include symptoms of paranoia that decrease the likelihood of using care. Additionally, some medications used to treat mental illnesses have side effects, such as insulin resistance, that can worsen comorbidities.
Graphic created for the blog from data in: Medford-Davis L, et al. The Role of Mental Health Disease in Potentially Preventable Hospitalizations: Findings From a Large State. Med Care. 2018;56(1):31-38.
Importantly, this new study illustrates the nuances of associations between mental illnesses and preventable hospitalizations. The figure above shows the odds ratios and 95% confidence intervals for the estimates from the paper’s multivariable analysis of factors associated with having any potentially preventable hospitalization (controlling for many other factors). All conditions but one (PTSD) were associated with increased odds of a potentially preventable condition. Breaking out the types of admissions, other nuances emerge. For example, bipolar disorder and schizophrenia increased the odds of most potentially preventable hospitalizations with notable exceptions for CHF, angina, and perforated appendicitis. Anxiety was associated with increased odds of hospitalization for angina, hypertension, asthma, and COPD. It was associated with decreased odds of admission for diabetes complications, UTI, and perforated appendicitis. Information on these differences could be very helpful to healthcare providers who have regular patients with these mental health conditions.
Of course, the analysis by Medford-Davis and colleagues cannot prove causation and did not account for many pre-hospital factors, previous treatment, and other social factors. The analysis was also unable to account for patients with multiple visits over time (data were de-identified) and so some of the standard errors might be smaller than they would be otherwise.
From other analyses, we know that potentially preventable admissions vary by region. Remote rural areas continue to have much higher rates than small metropolitan areas, and the difference is growing. Such information, when combined with other sources such as the Medford-Davis paper, suggests how we might identify specific barriers to appropriate care and which patients might need additional services to prevent future hospitalizations or readmissions. More research using data from additional states and other sources would be essential in this effort.
Jess Williams