Medicaid patients admitted to hospitals with a larger proportion of private payers received more diagnoses on Medicaid insurance claims than those in hospitals with a lower proportion of private payers, according to a new study of more than 1 million Medicaid admissions in New York state.
Diagnostic coding software is an investment in infrastructure that may be utilized more often by hospitals with a higher proportion of privately insured patients with higher reimbursement rates, wrote Kacie L. Dragan, MPH, a PhD candidate at Harvard University, Cambridge, Massachusetts, and colleagues.
“Provider-level variation in coding intensity has been documented to some degree for Medicare and commercially-insured groups, but little was known about diagnostic coding patterns for Medicaid-insured groups,” Dragan said in an interview.
“We also wanted to provide some evidence on the question of whether higher prices from private payers seem to incentivize hospital-level administrative investments, such as advanced EHRs or highly trained staff,” she explained. “If so, the impact of these administrative investments might spill over and be reflected in the number of diagnoses Medicaid patients receive.”
In a study published in JAMA Health Forum, Dragan and colleagues analyzed data from 1.6 million hospitalizations for Medicaid-insured patients between 2010 and 2017. The study population included Medicaid enrollees with at least two admissions in at least two different hospitals in New York state. The mean age of the patients was 48 years, 51.4% were women. Overall, 30.1% were White, 28.6% were Black, 23.3% were Hispanic, 4.6% were Asian, and 5.4% were other ethnicities.
Significantly more diagnoses were recorded when the same patient was seen in a hospital with more privately insured patients (0.03 diagnoses for each percentage point increase in the share of privately insured patients, P < .001).
Patients first discharged from hospitals in the bottom quartile of privately insured patients received 1.37 more diagnoses when subsequently discharged from hospitals in the top quartile, and those first discharged from hospitals in the top quartile of privately insured patients received 1.67 fewer diagnoses when discharged from hospitals in the bottom quartile (P < .001 for both).
Payment incentives appeared to play a role in the diagnostic codes used, the researchers noted. Diagnoses in hospitals with a higher share of private payers were significantly more likely to involve conditions sensitive to payment incentives, such as neuropathy or depression.
“The probability of receiving a commonly up-coded supplemental diagnosis increased by 2.50 percentage points when a Medicaid-insured patient was seen in a hospital with 40% privately insured patients compared with when they were seen in a hospital with just 10% privately insured patients,” the researchers write.
The results persisted in subgroup analyses and in a replication of the study using data from 2016 to 2017, after the implementation of the diagnostic code set ICD-10, with a similarly large increase of 0.06 additional diagnoses for each percentage point increase in the proportion of private paying patients.
The study findings were limited by several factors, including the use of only claims through 2014 in the main analysis, and the inability to determine whether patients are selecting into well-resourced hospitals for more complex conditions, the researchers noted. However, “To the extent that diagnoses drive reimbursement and quality scores, this may create a feedback loop that further benefits highly-reimbursed facilities and exacerbates inequity in resources,” the authors conclude.
“We were somewhat surprised to see such symmetry and a ‘dose-response’ gradient in the relationship between a hospital’s private payer share and the number of diagnoses coded,” Dragan told Medscape. Although many studies have focused on provider upcoding, “this finding may suggest that there could also be under-coding happening at the opposite extreme, among providers with large shares of Medicaid-insured patients; however, our study cannot say what the ideal level of diagnostic coding would be.”
Impact of Incentives Remain Unclear
“The diagnoses documented for Medicaid patients might, in part, be a reflection of the hospital’s payer mix and associated administrative style, rather than a reflection of a patient’s true underlying health,” Dragan said in an interview. “Disease surveillance metrics or risk-adjusted quality measures, for example, may be impacted by this variation in code capture, calling for caution when relying on patient diagnoses,” she added.
“Future research should aim to document whether this variation in diagnostic coding intensity has downstream implications for Medicaid patient treatment or outcomes,” said Dragan. “Additionally, it will be important to better understand what specific actions hospitals are taking in response to payer incentives, such as changing EHR [electronic health record] vendors or training staff, that might be behind this observed variation in coding intensity among Medicaid patients.”
Measure of Patient Risk is Needed
“Nearly all healthcare reforms require that we can accurately measure patient risk in order to compare providers or insurance plans,” Andrew Ryan, PhD, of the University of Michigan, Ann Arbor, said in an interview. “Factors other than true clinical severity that influence the measurement of patient risk, such as hospitals’ share of private patients, may result in inaccurate measurement.”
“This is a strong study,” said Ryan, a professor of health management at UM who was not involved in the research. “The authors found that a higher share of private patients led to greater risk coding for Medicaid patients; they attribute this effect to the fact that hospitals with a greater share of private (commercial) patients have stronger incentives to code,” he said.
However, “I’m not sure if this mechanism is driving the results,” Ryan noted. “For instance, I believe that the strongest incentives for upcoding risk is for Medicare Advantage patients, and these were categorized as public payers by the authors. Instead, I think that the likely mechanism for upcoding is that hospitals with more private patients are better resourced, and probably hired more coders.”
As for additional research, Ryan said he would be interested in seeing whether hospitals with more Medicare Advantage patients code more. “I would also be interested in understanding whether hospitals’ investments in coding staff drive the findings,” he said.
The study was supported by the Agency for Healthcare Research and Quality and the Commonwealth Fund. Dragan disclosed training fellowships from the Agency for Healthcare Research and Quality and from NIH’s National Institute of Mental Health. Ryan reports no relevant financial relationships.
JAMA Health Forum. Published online September 2, 2022. Full text
Heidi Splete is a freelance medical journalist with 20 years of experience.
For more news, follow Medscape on Facebook, Twitter, Instagram, YouTube, and LinkedIn
Source: Read Full Article