In an effort to improve the validity and person-centeredness of the nursing home resident assessment tool (the Minimum Data Set, or MDS), the Centers for Medicare and Medicaid Services introduced version 3.0 in October 2010. As a result, many of the measures and items health services researchers had grown accustomed to using in the MDS 2.0 changed — sometimes dramatically. While studies have explored these changes, validated new scales, proposed quality measures, and examined the reliability of the MDS 3.0, until recently we have not understood how changes in the reporting of active diagnoses in Section I of the MDS 3.0 affects analyses and interpretation of the MDS data.
The revised version of the MDS offered many improvements over the previous assessment tool, and some believe it has greatly improved patient care in the nursing home setting. Included in these changes were structured interviews with residents to assess their cognition, mood, pain, and preferences. In addition, the Confusion Assessment Method tool was included to screen for delirium, and revisions to the behaviors, psychosis, bladder, bowel, balance, falls, pressure ulcers and nutrition assessments, among many other changes, were made.
The active diagnoses recorded in the MDS 3.0 also underwent changes from its previous version. Specifically, the labels were updated and clarifications were added. Section I contains 56 “Active Diagnosis” checkbox options. These are designed to document diagnoses that have been present in the last 60 days and “have a direct relationship to the resident’s current functional status, cognitive status, mood or behavior status, medical treatments, nursing monitoring, or risk of death.” If a disease or condition is not specifically listed in Section I, the assessor is instructed to enter the diagnosis and ICD code in item I8000, which allows for 10 additional active diagnoses. The checkboxes are broadly labeled and no corresponding ICD codes are reported in the CMS’s Resident Assessment Instrument Version 3.0 Manual; therefore, it is possible that conditions may be captured only in the ICD fields and not in the corresponding diagnosis checkbox. Our team at Brown University set out to explore the diagnoses reported in Section I and understand the consistency of active diagnoses reported on the MDS assessment.
To do this, we used all new nursing home admission assessments in calendar years 2012 and 2013 and identified whether a condition was present in Section I, item I8000, using the ICD-9 codes for six common chronic conditions (i.e., Alzheimer’s disease, dementia, CVA/TIA/stroke, asthma/COPD/chronic lung disease, heart failure, and Parkinson’s disease).The team then investigated the rate by which individuals had these diagnoses present only in the MDS checkboxes (I0100-16500) or entered freely in item I8000. We found that 1 to 7% of MDS assessments had only the ICD-9 codes populated for the common chronic conditions we investigated. (Full results for each diagnosis can be found on our website, LTCfocUS.org) For more rare conditions (e.g., traumatic brain injury), we identified even greater rates of diagnoses listed only in the open coded section and not indicated in the corresponding check boxes.
Based on these findings, we strongly recommend that researchers working with the active diagnoses in the MDS 3.0 use the ICD-9 codes in item I8000 in addition to Section I’s checkboxes (items I0100-6500) when identifying active diagnoses among the population of nursing home residents. Without this, it is likely a number of residents with diagnoses of interest will be missed.