The Usefulness of Present-on-Admission Data as an Indicator of Healthcare Quality Evaluation Using the Korean National Hospital Discharge in-Depth Injury Survey Data from 2006 to 2019

Introduction

Comorbidities affect the treatment and outcomes of an illness.^1–3 Comorbidities can be divided into underlying diseases that existed before the onset of the principal diagnosis and before complications that occur during treatment at the hospital.

Comorbidities of the principal diagnosis may have varying effects on the disease and require different management depending on the onset timing. Cardiovascular comorbidities of the coronavirus disease 2019 (COVID-19), which has emerged as an important issue, manifest and are treated differently depending on whether they were an underlying disease before the COVID-19 infection or whether they were a complication of COVID-19.^4,5 In particular, complications that occur during the process of treating the principal diagnosis are subject to healthcare quality evaluation and cause analysis. Thus, present-on-admission (POA) indicators and Patient Safety Indicators (PSI) have been implemented to manage the onset timing of comorbidities and patient safety.

PSI were developed by the Agency for Healthcare Research and Quality (AHRQ); they are used to identify and manage diseases that are likely to occur during the process of treating patients at a healthcare facility and threaten patient safety.⁶ There are 20 AHRQ PSI,⁷ and the indicators are continually reviewed by the Organisation for Economic Co-operation and Development (OECD).^8,9 The POA indicator is a system for distinguishing between a new diagnosis that was made during a hospital stay (POA) and one that was present before admission (POA Y flag) when healthcare facilities bill for reimbursement. The US Centers for Medicare & Medicaid Services (CMS) requires all Medicare and Medicaid claims to include the POA indicator in the principal and secondary diagnoses since 2007. From 2008, the CMS makes reimbursement decisions for diseases defined as a hospital-acquired condition (HAC) based on the POA indicator on the patient’s discharge data and declines reimbursement of 10 types of HAC based on the POA indicator.^10,11

Studies that examined the effects of implementing the POA indicator in Medicare and Medicaid reimbursement denials for HACs and pressure ulcers^12,13 showed that reimbursement was reduced, further highlighting the importance of preventing and managing HACs.

In 2012, the diagnosis-related group (DRG) reimbursement model was implemented in South Korea. Healthcare facilities are now required to specify the POA indicator in the DRG and new DRG reimbursement claims.¹⁴ The new DRG reimbursement system, which is an expansion of the previous seven systems, currently runs as a pilot program.¹⁵ In addition, the Health Insurance Review and Assessment Service (HIRA) provides subsidies for participating healthcare facilities to facilitate the establishment of the new DRG system and periodically performs qualitative evaluations of medical records, diagnostic coding, and POA data.¹⁶

POA data in Korea contribute to the thoroughness of medical record documentation and the enhanced accuracy of diagnostic coding.¹⁷ However, data utility remains low due to poor accuracy resulting from the short time elapsed since implementation and relevant staff and healthcare professionals’ inadequate awareness and training.^{10,14,16,18–20}

In other countries, studies examining the conditions that markedly impact severity or treatment process and outcomes, such as hemorrhage, thrombosis, anemia, and coronary artery disease, have utilized the POA indicator to analyze the cause of HACs and promote quality of care.^21–25 For instance, in the United States, POA data flagged in Medicare and Medicaid claims are utilized to monitor the quality of care provided to low-income individuals and older adults.^26,27

POA data—along with PSI—may be an important source of information for detecting HACs, and it serves as a useful indicator to identify and ameliorate the risk factors and causes of complications. Hence, it is strongly associated with the quality of diagnostic coding^28–31 and ultimately is a crucial means to promote the quality of healthcare.

However, a study that investigated the accuracy of POA data collected since 2012 in Korea based on PSI reported that the data has low reliability and accuracy and thus cannot be used as an indicator of healthcare quality. The study also suggested that diagnoses flagged with an N POA indicator, as opposed to the admitting diagnosis, should be monitored.¹⁶

Thus, this study aims to investigate whether decubitus ulcers, delirium, and hypokalemia—three conditions ranked in the top 20 diagnoses flagged with an N POA indicator in the new DRG claims data¹⁰ are useful as indicators of healthcare quality. In particular, POA codes for decubitus ulcers and delirium have been reported to have significant benefits in patient treatment and statistics data.^32,33 Further, because the ICD-10 codes for these conditions do not indicate whether the condition was hospital-acquired or is a complication of treatment (T code), the POA flag is particularly useful in distinguishing this. Hence, if they are useful indicators of healthcare quality, our findings will provide evidence supporting their use as target diseases for POA management.

Materials and Methods

This study analyzed patient discharge data for a 14-year period from 2006–2019 using the Korean National Hospital Discharge In-Depth Injury Survey (KNHDIS). The KNHDIS aims to produce statistics based on the scale and epidemiological characteristics of injuries in Korea and to provide basic data for the development of healthcare policies to promote public health. To achieve this goal, the survey recruits all patients who have been discharged from a 100-bed or larger hospital nationwide each year as its study population and uses a two-step stratified cluster sampling to select about 9% of patients who were discharged from approximately 200 sampled hospitals. The survey is composed of 20 general items, such as disease and treatment information, and 10 in-depth injury items, including intentionality of injury, mechanism, place of occurrence, and activity, based on medical records. In this study, we focused on only the general items.

After examining the characteristics of all discharge patients within a 14-year period, those with a secondary diagnosis code (ICD) of L89 (decubitus ulcer), F058 or F059 (delirium), or E876 (hypokalemia) were identified and classified. Discharge patients who had more than one of the three secondary codes were excluded from the analysis. The three secondary code groups were compared for patient’s sex (men, women), age (<65, ≥65), route of admission (emergency department, outpatient department, others), type of insurance (National Health Insurance, medicaid, others), hospital bed size (100–299, 300–499, 500–999, ≥1000), treatment outcomes (alive, death), surgery (no, yes), and length of stay (LOS) using chi-square tests and one-way ANOVA.

The frequency of principal diagnoses of the three secondary diagnosis code groups (decubitus ulcers, delirium, hypokalemia) was examined based on the third digit of the ICD-10 code. Subsequently, the 20 most frequent principal diagnoses were identified. For patients with one of the top 20 principal diagnoses, the associations between their secondary diagnosis (decubitus ulcers, delirium, hypokalemia) and surgery, death, and LOS were analyzed using logistic regression and linear regression. The results were adjusted for sex, age, route of admission, type of insurance and hospital bed size. An adjusted odds ratio (aOR) with a 95% confidence interval (CI) was presented for the logistic regression, and an unstandardized coefficient and standardized coefficient were presented for the linear regression.

Statistical analyses were performed using the SAS version 9.4 software, and p<0.05 was deemed statistically significant.

Results

Sociodemographic Characteristics of the Study Population

From 2006 to 2019, 49.3% of all discharge patients were male, and 50.7% were female. The mean age was 46.2 years old (SD 25.2). Of all the discharged patients, 1.8% died and 29.8% had surgery. The mean LOS was 8.5 days (SD 24.0) (Table 1).

Table 1 Sociodemographic Characteristics of the Total Study Population and Patients with a Secondary Diagnosis of Decubitus Ulcers, Delirium, or Hypokalemia

Sociodemographic Characteristics of Patients with a Secondary Diagnosis of Decubitus Ulcers, Delirium, or Hypokalemia

Of 19,871 discharge patients, 10,390 (52.3%) patients had a secondary diagnosis of decubitus ulcers, 6103 (30.7%) had delirium, and 3378 (17.0%) had hypokalemia (Table 1).

Regarding gender, there was a higher percentage of males in the decubitus ulcer group (male 54.6%, female 45.4%) and the delirium group (male 56.1%, female 43.9%), but there was a higher percentage of females in the hypokalemia group (female 63.8%, male 36.2%) (p<0.0001).

The mean age was 70.1 years (±16.1) in the decubitus ulcer group, 73.8 years (±13.3) in the delirium group, and 64.4 years (±21.0) in the hypokalemia group. In all three groups, there were more older adults aged 65 years and over than those under 65 years, but the percentage of individuals under 65 years was higher in the hypokalaemia group compared to the other groups (p<0.0001).

In contrast to the general population, there were more patients admitted through the emergency department than through outpatient services in all three groups (p<0.0001), and national health insurance was the more common type of health insurance over medical aid (p<0.0001).

Regarding treatment outcomes, more people survived than those who died, but the percentage of deaths was higher in the decubitus ulcer group compared to the other two groups (p<0.0001).

The rate of surgery was the highest in the delirium group (p<0.0001), and LOS was the highest in the decubitus ulcer group (45.5±92.1 days), followed by the delirium group (27.9±36.9 days) and the hypokalemia group (18.7±28.4 days) (p<0.0001).

Distribution of Principal Diagnosis and Associations in the Decubitus Ulcer Group

The most frequently occurring principal diagnoses in the group of patients with a secondary diagnosis of decubitus ulcer was J18 (pneumonia, organism unspecified) (n=922, 8.87%), followed by J69 (pneumonitis due to solids and liquids) (n=437, 4.21%), I63 (cerebral infarction) (n=378, 3.64%), N39 (other disorders of the urinary system) (n=332, 3.20%), and A41 (other sepsis) (n=305, 2.94%). Approximately 47.93% of the patients in this group had one of the top 20 principal diagnoses (Table 2).

Table 2 Top 20 Principal Diagnoses in the Group with a Secondary Diagnosis of Decubitus Ulcer and the Association Between a Decubitus Ulcer and Treatment Outcomes

Regarding the association between a secondary diagnosis of decubitus ulcers and surgery among patients with one of the top 20 principal diagnoses, those who had a secondary diagnosis of decubitus ulcers had 1.566 to 7.876 times higher odds of surgery than those without a secondary diagnosis of decubitus ulcers (p<0.0001).

Regarding mortality risk, patients with one of the top 20 principal diagnoses (excluding G82 and I61) and a secondary diagnosis of decubitus ulcers had 1.523 to 39.085 times higher mortality risk than those without a secondary diagnosis of decubitus ulcers (p<0.0001).

Regarding LOS, patients with one of the top 20 principal diagnoses and a secondary diagnosis of decubitus ulcers had significantly longer LOS than those without a secondary diagnosis of decubitus ulcers (p<0.0001).

Distribution of Principal Diagnosis and Associations in the Delirium Group

The most frequently occurring principal diagnoses in the group of patients with a secondary diagnosis of delirium was S72 (fracture of femur) (n=425, 6.96%), followed by J18 (pneumonia, organism unspecified) (n=282, 4.62%), C34 (malignant neoplasm of bronchus and lung) (n=220, 3.60%), S06 (intracranial injury) (n=212, 3.47%), and I63 (cerebral infarction) (n=175, 2.87%). Approximately 44.78% of the patients in this group had one of the top 20 principal diagnoses (Table 3).

Table 3 Top 20 Principal Diagnoses in the Group with a Secondary Diagnosis of Delirium and the Association Between Delirium and Treatment Outcomes

Regarding the association between a secondary diagnosis of delirium and surgery among patients with one of the top 20 principal diagnoses, those with a secondary diagnosis of delirium had 1.633 to 3.679 times higher odds of surgery than those without a secondary diagnosis of delirium (p<0.0001).

Regarding mortality risk, patients with a principal diagnosis of C34, Z51, C22, or C16 and a secondary diagnosis of delirium had 2.935 to 16.991 times higher mortality risk than those without a secondary diagnosis of delirium. However, patients with a principal diagnosis of S06 or I61 had a lower mortality risk when they had a secondary diagnosis of delirium (p<0.0001).

Regarding LOS, patients with one of the top 20 principal diagnoses (excluding J69) and a secondary diagnosis of delirium had a significantly longer LOS than those without a secondary diagnosis of delirium (p<0.0001).

Distribution of Principal Diagnosis and Associations in the Hypokalemia Group

The most frequently occurring principal diagnoses in the group of patients with a secondary diagnosis of hypokalemia was J18 (n=244, 7.22%), followed by E87 (n=167, 4.94%), A09 (n=154, 4.56%), K70 (n=88, 2.61%), N39 (n=88, 2.61%), and N10 (n=84, 2.49%). Approximately 44.94% of the patients in this group had one of the top 20 principal diagnoses (Table 4).

Table 4 Top 20 Principal Diagnoses in the Group with a Secondary Diagnosis of Hypokalemia and the Association Between Hypokalemia and Treatment Outcomes

Regarding the association between a secondary diagnosis of hypokalemia and surgery, patients with a principal diagnosis of Z51 and S06 with a secondary diagnosis of hypokalemia had 7.116 times higher odds and 2.168 times higher odds of surgery, respectively, compared to those without a secondary diagnosis of hypokalemia. However, patients with a principal diagnosis of N39 or N18 had lower odds of surgery when they had a secondary diagnosis of hypokalemia (p<0.0001).

Regarding mortality risk, patients with a principal diagnosis of J18, A09, N39, Z51, S06, K56 or D34 and a secondary diagnosis of hypokalemia had 2.204–9.574 times higher mortality risk than those without a secondary diagnosis of hypokalemia (p<0.0001).

Among patients with a principal diagnosis of J18, A09, N39, N10, I63, E11, J15, J44, Z51, K56 and D34, those with a secondary diagnosis of hypokalemia had a significantly longer LOS than those without a secondary diagnosis of hypokalemia (p<0.0001).

Discussion

In this study, we chose delirium (F058, F059), decubitus ulcer (L89), and hypokalemia (E876) as diseases of interest, as they are three of the most common conditions with an N POA indicator in the new DRG, according to a previous study.¹⁰ We used the KNHDIS data for a 14-year period between 2006 and 2019 to identify the top 20 principal diagnoses with a secondary diagnosis code for one of the three conditions and compared death, surgery, and LOS according to the presence of these secondary diagnosis codes to explore the usefulness of these conditions as the targets of POA management. Comorbidities impact the severity and treatment outcomes of the principal diagnosis¹ and the treatment process and required resources vary depending on the timing of their onset.³⁴ Thus, diseases that influence treatment outcomes but are not present on admission must be managed as healthcare quality indicators.

The sociodemographic characteristics of patients who had additional diagnoses of decubitus ulcers, delirium, and hypokalemia were examined, and all three groups had a high proportion of patients aged 65 or older. This is attributable to the fact that these conditions are caused by chronic diseases prevalent among older adults,³⁵ and delirium is a common postoperative complication in older patients.^36,37 Therefore, intensive care with varying management timing is needed for older adults depending on their POA indicator. Furthermore, the rate of death was higher among patients with decubitus ulcers than those with delirium or hypokalemia. Decubitus ulcers tend to occur more in socially and economically vulnerable patients^38,39 and they prolong the duration of treatment and increase the risk for sepsis, thereby elevating overall risk.^40,41 Thus, decubitus ulcers require thorough management from the time of admission (if present on admission), and, if not present on admission, intensive nursing care and intervention protocol should be followed with patient monitoring to prevent this complication during hospital stays.^42,43

Decubitus ulcers are defined as local injuries caused by disrupted blood flow to the skin or subcutaneous tissue as a result of continuous pressure on bony prominences.⁴⁴ In particular, the prevalence of decubitus ulcers is rising owing to the increasing older adult population and prevalence of chronic diseases, consequently increasing relevant treatment cost.⁴⁵ Accordingly, the need for a differentiated management system for decubitus ulcers according to onset timing before and after hospitalization has been emphasized.^46,47 In this study, the differences comparison in the distribution of principal diagnoses between groups with and without decubitus ulcers as a secondary diagnosis showed that the group with decubitus ulcers as a secondary diagnosis had a higher prevalence of chronic diseases such as cerebral infarction, brain injury, and chronic renal disease. They also had 1.566 to 7.876 times higher odds of surgery and 1.523 to 39.085 times greater mortality risk compared to the group without decubitus ulcers. Additionally, LOS increased significantly in all of the patients with one of the top 20 principal diagnoses. This suggests that decubitus ulcers are a meaningful indicator of patient safety and healthcare quality and are a viable target disease for POA management.^48–50

The most common principal diagnosis for patients with delirium as a secondary diagnosis was surgery due to femoral fracture (6.96%). This is similar to previous results, which show that delirium is more common in postoperative older adult patients and that femoral fracture is a common type of fracture among older adults.^37,51,52 Prompt treatment is crucial to prevent or minimize the adverse consequences of postoperative delirium, which is a form of cerebral dysfunction that occurs postoperatively.⁵³ Therefore, delirium is an important target disease for POA management and is a useful patient safety and healthcare quality indicator, especially for older adult surgical patients. Additionally, we compared the mortality rates between the groups with and without delirium as a secondary diagnosis among those with one of the top 20 principal diagnoses and observed that patients with lung cancer had a 2.935 to 16.991 times higher mortality risk and longer LOS when they had delirium as a secondary diagnosis. This supports previous findings that delirium is an indicator requiring management.^54,55

Potassium is the most abundant electrolyte inside cells, and hypokalemia refers to a state in which the serum potassium level drops below normal. Potassium must be consumed through food, and since it is abundant in the gastrointestinal tract, hypokalemia can occur from malnutrition, osmotic diuresis, insulin excess, and gastrointestinal fluid loss caused by vomiting and diarrhea. It is often asymptomatic but can affect cardiac function, making its diagnosis and treatment important.⁵⁶ Further, hypokalemia is common postoperatively, particularly as a common complication following total thyroidectomy.⁵³ In this study, the top 20 principal diagnoses among patients with a secondary diagnosis of hypokalemia included gastrointestinal disorders and thyroid cancer. In this group, the odds of surgery, LOS, and mortality risk differed according to the presence of hypokalemia as a secondary diagnosis, with the hypokalemia group having 2.204 to 9.574 times higher mortality risk than those without hypokalemia. Given that hypokalemia impacts the severity of the patient’s condition and frequently occurs postoperatively, that is, during the treatment process, hypokalemia should be managed as a target for POA management and as a healthcare quality indicator.⁵⁷

The results of this study demonstrated that treatment outcomes, odds of surgery, and LOS differed according to the presence of decubitus ulcers, delirium, or hypokalemia as a secondary diagnosis flagged with an N POA indicator among patients with one of the top 20 principal diagnoses. The results highlighted that decubitus ulcers, delirium, and hypokalemia are significant target diseases for POA management. However, we could not monitor the accuracy of disease coding for discharged patients reported in the KNHDIS data owing to the extensive dataset spanning 16 years. Additionally, we could not comparatively analyze the diseases of interest according to the N POA indicator, that is, the timing of onset, owing to the limited availability of study data. Therefore, we investigated the usefulness of these diseases as targets of POA management and healthcare quality indicators by comparing the differences according to the presence of these diseases as a secondary diagnosis.

Conclusion

Our findings showed that decubitus ulcers, delirium, and hypokalemia are common conditions that arise postoperatively or during hospital stays. As such, patients with one of the top 20 principal diagnoses have increased mortality risk, surgery risk, and LOS when one of these three conditions is a secondary diagnosis. As decubitus ulcers, delirium, and hypokalemia are linked to the severity of the principal diagnosis, these conditions should be designated as target diseases for POA management for monitoring. Ultimately, they will be useful as healthcare quality indicators. Our findings provide useful foundational data for selecting target diseases for POA monitoring and healthcare quality indicators.