Diagnosis-Related Group Weight and Derived Case Mix Index to Assess the Complexity among Twins

Abstract

Objective

Diagnosis-related groups (DRG) are used to summarize hospital morbidity and mortality. Each DRG has a weight which is important in calculating the case mix index (CMI), a numeric summary of disease complexity in a population of patients. We utilized DRG weight and resultant CMI to compare postnatal outcomes among singletons versus monochorionic and monoamniotic, monochorionic diamniotic, and dichorionic diamniotic twins.

Study Design

This single-center and retrospective cohort study evaluated DRGs assigned by the investigators, birth weight, gestational age, length of stay (LOS), NICU admission rate, and mortality in twin births between 2014 and 2016. Twins were analyzed depending on chorionicity and amnionicity. Overall, 3 months of singleton births served as the control. The CMI derived from DRG weights were compared across groups.

Results

Twins (n = 288) had lower gestational ages and birth weights and higher mortality, LOS, NICU admission rates and DRG weights/CMI compared with singletons (n = 327; p < 0.001 for each). The LOS was no different between twin subtypes; monochorionic monoamniotic twins had the highest mortality and DRG weight (p < 0.001).

Conclusion

DRG weight and CMI values summarize in-hospital complexity and can be a useful tool to evaluate differences in care complexity among groups of patients.

The diagnosis-related group (DRG) is a Medicare payment designation that codifies the complexity of a hospitalized patient’s care and the expected amount of resources needed for that patient’s hospital stay. The DRG weight is a numeric interpretation of that complexity and is ultimately factored into hospital payments.¹ DRG and DRG weights are typically determined on the patient’s day of discharge, retrospectively evaluating patients for morbidities and mortality. The DRG weight is used to calculate the case mix index (CMI) of a specific care unit in a hospital. The CMI is a factor that is important for payment from Medicare and Medicaid services. The Center for Medicare and Medicaid Services (CMS) in the U.S. Department of Health and Human Services (DHHS) recognizes that although DRGs are used primarily for payment purposes, clinicians can utilize the DRG to describe severity of illness, treatment difficulty, and prognosis in patients within a given clinical group.² While this additional purpose is emphasized, an application of DRGs and CMI in this manner has not been described in the literature. The DRG weights and CMI values are used to compare neonatal intensive care unit (NICU) patients among different centers by collaborative groups such as the American Association of Medical Colleges or with the Vizient group. This aspect of the DRG and CMI has been underutilized by clinicians.

Clinicians may be unfamiliar with the use of DRG, DRG weight, and resultant CMI. For a patient older than 28 days, the assignment of a proper DRG can be a daunting endeavor with 740 different DRGs plus designation of significant or major comorbidities, resulting in over 2,000 choices. For newborns, there are only seven DRGs with their corresponding weights, making it relatively easy to navigate (Table 1). The DRG weight ranges from 0.17 for a normal newborn to 5.26 for those infants who were extremely premature (defined as either ≥26 weeks’ gestation at birth or ≥999 g birth weight) or diagnosed with respiratory distress syndrome (RDS). Besides the DRG for RDS and/or extreme immaturity, an infant may fit into one of three term categories (with minimal comorbidities [795], significant comorbidities [794], or major comorbidities [793]). Normal newborns can include patients with minor complications such as cephalhematoma, delivery by breech presentation or with forceps, jaundice not otherwise specified, feeding difficulties, and liveborn twin deliveries by cesarean or vaginal routes. There are two preterm categories defined as 27 to 36 weeks’ gestation inclusively (with significant comorbidities [792] or with major comorbidities [791]). Finally, there is a DRG attached to infants that die or are transferred to another institution within the first several days of life (789). Examples of major comorbidities for neonates include neonatal or perinatal respiratory failure, any intraventricular hemorrhage, retinopathy of prematurity, or pneumothorax. Examples of significant comorbidities include small or large for gestational age (GA) status, hypoglycemia in an infant of a diabetic mother, newborns affected by any maternal comorbidity, or mild hypoxic ischemic encephalopathy. Table 2 provides a more detailed listing of significant and major comorbidities used to determine an appropriate DRG. With the simplicity of neonatal DRGs, it is possible to accurately assign the correct DRG in real time, often within the first or second day of life. For a given population, the average weight of all patients is the CMI.

Table 1.

Neonatal diagnosis-related groups and corresponding diagnosis-related groups weights

789	Neonates died or transferred to another acute care facility	1.5979
790	Extreme immaturity or respiratory distress syndrome, neonate	5.2692
791	Prematurity with major problems	3.5987
792	Prematurity without major problems	2.1713
793	Full-term neonate with major problems	3.6967
794	Neonate with other significant problems	1.3084
795	Normal newborn	0.1771

Notes: Extreme immaturity is defined as either ≥26 weeks’ gestation at birth or ≥999 g birth weight. Prematurity is defined as 27 to 36 weeks’ gestation inclusively. Normal newborns can include patients with minor complications such as cephalhematoma, delivery by breech presentation or with forceps, jaundice not otherwise specified, feeding difficulties, and liveborn twin deliveries by cesarean or vaginal routes. These categories were developed in 1983 along with incorporation of the entire diagnosis-related groups (DRG) system by Medicare. The DRG weights are representative of the average resources required to care for a patient within a particular DRG compared with the average resources needed to treat patients in all other DRGs.²

For purposes of this project, since death would be considered the most significant comorbidity, an extra weight of 1.5979 (DRG of 789 infants transferred or died) could be added to the final DRG weight for any neonate who died.

Table 2.

Selected examples of comorbidities affecting neonatal diagnosis-related groups within the study cohort

Significant comorbidities

Newborn affected by maternal conditions such as chorioamnionitis, hypertensive disorders, placental transfusion syndromes, oligohydramnios, polyhydramnios, premature and prolonged rupture of membranes, or tobacco use, fetal drug exposure, and maternal infection. Respiratory distress of the newborn not otherwise specified transient tachypnea of the newborn Typical triggers for neonatal sepsis evaluation (diagnoses: metabolic acidemia, cyanotic attacks of the newborn, other apnea of the newborn, neonatal tachycardia, neonatal bradycardia, transitory ileus of the newborn, and disturbances of temperature regulation of newborn) Meconium staining at delivery Large for gestational age or small for gestational age Syndrome of an infant of diabetic mother (gestational diabetes or other) Mild neonatal encephalopathy

Major comorbidities

Any electrolyte abnormality Neonatal hypovolemia Bacteremia, sepsis, pneumonia, urinary tract infection, or meningitis, including culture negative sepsis (clinical infection) Hydrocephalus or congenital malformation of the brain or nervous system Meconium aspiration (with or without respiratory symptoms) Moderate/severe neonatal encephalopathy Neonatal hypoglycemia (without maternal diabetes) Small for gestational age (<2,500 g) with evidence of poor fetal nutrition Congenital small head/microcephaly Neonatal withdrawal symptoms from maternal use of drugs of addiction Neonatal pneumothorax or pneumomediastinum Primary atelectasis of the newborn Perinatal/neonatal respiratory failure Ventricular septal defect Retinopathy of prematurity (any grade) Intraventricular hemorrhage (any grade)

Note: In our study cohort, the above major and significant problems most commonly determined the final diagnosis-related groups (DRG). Related diagnoses have been grouped. A complete listing of major and significant problems for DRG weight determination can be found through the U.S. Department of Health and Human Services website.² Some major problems, such as transfusion reactions, fractures, and complications of foreign bodies, are not typically seen in the neonatal population.

Since the DRG weight reflects various prenatal and postnatal complications for NICU patients, it could be used to summarize the severity of neonatal illness, with more detail than the GA at birth alone provides. The CMI from entire NICU services are also used to compare NICU patients’ complexities among different centers by collaborative groups as noted previously, we hypothesized that the CMI would be useful tools to evaluate differences in care complexity among groups of patients at risk as suggested by CMS.² Since death would be considered the most significant comorbidity, an extra weight of 1.6 was added as derived from DRG (789): infant transferred or died could be added to the final DRG weight for any neonate who dies. With this manipulation, we constructed a neonatal morbidity/mortality graph (Fig. 1). The graph ranges from normal term newborn with expected 0% mortality (weight = 0.17) to a 22-week preterm with 49.2% mortality based on 2019 VON data (weight = 6.38).³ Using these same calculations, we calculated the same for inborn singletons from a high-risk delivery service (mother/baby care), term infant with significant congenital heart disease,⁴ and singleton infants admitted to our own level 4 NICU.

Fig. 1.

Graph depicting CMI weight of normal newborns, newborns from high risk OB service, term newborns with significant CHD and infants born at 22 weeks. CHD, congenital heart disease; CMI, case mix index.

To further investigate the utility of this DRG/CMI, we chose to apply it to evaluate short-term outcomes of the different subsets of twins versus singletons.

Importance of multiple births: Multiple gestations have increased in frequency in the United States since 1980.⁵ The steady increase in multiple births has been associated with the use of fertility therapies and advancing maternal childbearing age.^6,7 In 2014, the CDC National Vital Statistics described a peak in the U.S. rate of twinning at 33.9 per 1,000 births.⁵ Twins and higher order multiples have increased health care utilization in the neonatal period due to higher rates of NICU admission, preterm birth, and growth restriction.⁸ Adverse outcomes are more frequent in monozygotic pregnancies due to fetal complications such as twin–twin transfusion syndrome and intrauterine growth restriction as shown in a large Danish cohort,⁹ though dizygotic pregnancies are not without risks.¹⁰ Our project attempts to predict a level of complexity (CMI) for a neonatal patient population by matching documented comorbidities to the corresponding DRG and using the DRG weights to differentiate outcomes of twin pregnancies based on zygosity and chorionicity.

Materials and Methods

This retrospective cohort study was performed at a single, level IV, academic NICU with an average of 1,700 deliveries annually over the 3-year period of review,¹¹ with approximately 70% of these deliveries classified as high risk. During the period of review, the resuscitation was offered at our institution to neonates born at 23 weeks’ gestation or higher, with those at the limits of viability being resuscitated per parent wishes. We reviewed the electronic medical records of all twin gestations with at least one live born infant from June 2014 to June 2016 (144 twin pairs). For a control group, we evaluated all inborn singletons born during the month of June from the three consecutive years of study (2014, 2015, and 2016; 327 infants). Triplets and higher order multiples were excluded from review as evaluating chorionicity and amnionicity for higher order multiples becomes more complex. With the exception of higher order multiples, there were no other exclusions for our evaluations. The West Virginia University Institutional Review Board approved the study protocol.

• Measures (birth group): Cases were sorted into groups based on singleton or twin gestation groups. Twins were differentiated as monochorionic-monoamniotic (MM), monochorionic-diamniotic (MD) or dichorionic-diamniotic (DD) gestations.

• Individual patient DRG weight: The method to determine appropriate DRG and weight is shown in Tables 1 and 2, based on the CMS evaluation tools. A complete listing of major and significant problems for DRG weight determination can be found through the U.S. Department of Health and Human Services website.² Some major problems, such as transfusion reactions, fractures, and complications of foreign bodies, are not typically seen in the neonatal population. Two investigators (R.J. and M.P.) independently determined each patient’s DRG, and resultant weight based on this system by reviewing the problem list in the electronic medical record (EPIC Systems, Verona, WI). As noted earlier, an extra weight of 1.5979 (DRG of 789 infants transferred or died) was added to the final DRG weight for those neonates who died. The CMI is an average DRG weight of the different populations of infants.

• Birth weight: The birth weights of twins and singletons were recorded in kilograms.

• Gestational age: The GAwas defined as the last completed week of gestation based on the best obstetric estimate, such that 23^6/7 weeks was counted as 23 completed weeks.

• NICU admission: After delivery, neonates at our institution are either admitted to the newborn nursery for rooming-in with mother or admitted to the NICU if additional monitoring or care is required. Staying less than 24 hours is still considered NICU admission.

• Length of stay: The LOS was defined as the number of days until discharge or death.

• Mortality: Mortality rate was computed based on the number of deaths occurring during the NICU stay divided by the number of live born neonates from each specific subgroup. (singleton, DD, MD, and MM).

Statistical Analysis

Descriptive analyses were conducted to obtain means, standard deviations, and other measures about the variable ranges and characteristics in this sample. Group differences in birth weight, GA, LOS, and mortality were examined by using one-way analyses of variance (ANOVAs), where group (singleton or subtypes of twin) was the independent variable. A Chi-square analysis was used to assess any differences between the groups based on singleton, MM, and DD groups. To examine differences in these outcomes based on DRG weight, we calculated another MANOVA with DRG weight as the independent variable. Bonferroni post hoc analyses were used to examine differences within the subcategories of multiples. All analyses were implemented by using SPSS software version 26.0 (IBM Corp., Armonk, NY).

Case Mix Index Graph

The CMI from each twin group was plotted onto the CMI graph as a visual tool with respect to morbidities of the twin groups versus the wide range between normal newborn, extreme immature, and critically ill newborn (Figs. 1 and 2).

Fig. 2.

Addition to CMI weight graph from Fig. 1 with MM, MF, and DD twins and singleton infants admitted to NICU from the study period. CHD, congenital heart disease; CMI, case mix index; MD, mono-mono; MF, mono-di; DD, di-di twins; NICU, neonatal intensive care unit.

Results

Baseline demographics:

A total of 615 infants were evaluated, with 288 twins (144 twin pairs), and 327 singletons in the sample. Among the twins, 5.5% were MM, 21.5% were MD, and 73% were DD.

All infants in the study ranged from 23 to 41 completed weeks of gestation at birth. As shown in Table 3, all twins were born at a significantly younger mean gestation than singletons (F = 129; df = 3, p < 0.01) based on our calculated univariate analyses. The MM twins were born significantly earlier than MD or DD twins (p < 0.01). Each group of twins had on average, significantly lower birth weights (F = 5.6; df = 2; p < 0.01) than the mean of 3.1 kg observed in singletons. Within the twin subgroups, MM were significantly lower in birth weight compared with DD twins (p < 0.03), but not MD twins.

Table 3.

Mean comparisons of postnatal outcomes by pregnancy type

Pregnancy type	Mean GA at Birth in weeks ± SD	Mean birth weight in kg ± SD	NICU admission rate (%)	Mean LOS in days ± SD	Mean CMI ± SD	Mortality rate (%)
All singletons (n = 327)	37 ± 2.8a	3.1 ± 0.8c	22e	7.3 ± 19.9g	1.7 ± 1.6h	1.5k
All twin gestations (n = 288)	33 ± 2.9a	1. ± 0.7c	83e	24.7 ± 17.9g	3.8 ± 1.6h	6.9k
DD twins (n = 210)	33 ± 3.4b	1.9 ± 0.6c, d	81f	24.3 ± 35.3	3.6 ± 1.6i	3.3j
MD twins (n = 62)	32 ± 3.7b	1.8 ± 0.7c	69f	25.9 ± 34.9	3.8 ± 1.6i	12.9j
MM twins (n = 16)	29 ± 2.9b	1.4 ± 0.7c, d	100f	27.2 ± 17.8	5.1 ± 1.6i	37.5j

Abbreviations: CMI, case mix index; DD, dichorionic diamniotic; GA, gestational age; LOS, length of stay; MD, monochorionic diamniotic; MM, monochorionic monoamniotic; NICU, neonatal intensive care unit; SD, standard deviation.

Notes: Univariate analyses of variance were conducted to compare average gestational age, birth weight, neonatal intensive care unit admission rate, length of stay, case mix index, and mortality rates across singletons, twin gestations, and subtypes of twin gestations.

^{a,b,c,e,g,i,j}Comparisons marked with the following superscripts were significant at the p <0.01 level

^d,fp <0.05 level; and

^h,kp <0.001 level.

Group comparisons (singleton, DD, MD, and MM) based on NICU admission rate and hospital LOS: Our univariate calculations also revealed that among singletons, 22% were admitted to the NICU during the study period. NICU admission occurred significantly more often in twins (p < 0.001), and in all of the MM twins. MD twins had significantly fewer NICU admissions that other twin types (p < 0.05). All twins had significantly longer hospital stays compared with singletons (F = 20.1; df = 3; p < 0.01). There was no difference in the LOS among the three twin subgroups.

NICU admissions:

Singletons from the study period who were admitted to the NICU were separately evaluated in a post hoc subanalysis. Their mean GA was 34 weeks at birth, mean birth weight of 2.27 kg, average LOS was 23 days, mortality rate 5%, and mean CMI of 4.2. The MM twins have a significantly higher CMI than those singletons requiring NICU admission (p < 0.05). There was no significant difference in the complexity of care as determined by CMI in singleton NICU admissions versus MD or DD twins.

CMI and mortality rate for all groups are detailed in Table 3: The CMI was significantly higher in all subtypes of twins compared with all singletons (p < 0.001). Being a twin itself was not factored into the individual DRG weight. There was again a significant difference between MM and other twin types. The mortality rate was higher among MM twins when compared with other groups. Five out of the six deaths in MM twins occurred within the first week of life. Only two of the MM twin deaths occurred due to extreme prematurity. The majority of the MD twin deaths (five-eighths) occurred due to complications of either extreme prematurity or twin–twin transfusion syndrome. The results were then plotted onto our CMI morbidity/mortality graph. (Fig. 2). The CMI graph ranges from total normal newborn with CMI of 0.17 to a CMI of 6.38 for a 22-week-old infant. On the CMI graph, DD and MD twin sets have a CMI that is higher than a singleton delivered at the same nursery, but lower than score for singleton admitted to the NICU. Our MM twin sets showed much higher CMI likely related to higher percentage of infants meeting DGR 790 criteria (RDS or less than 26 weeks’ gestation at birth) and higher percentage of these infants dying.

Effects of DRG on study outcomes:

We noted a significant main effect of the DRG weight variable when examining some of the outcomes in this study (F = 83.02; df = 3; p < 0.001) when exploring associations and the specific DRG role. Specifically, the DRG weight variable was significantly associated and helped distinguish between singleton, MD, and DD infants. As noted earlier, DRG weight was lowest among singleton births and highest among MM infants (p < 0.001). The DRG weight variable was associatedwith differences in birth weight (p < 0.001), GA (p < 0.001), and LOS (p < 0.001).

Discussion

For the adult population, determining a DRG and its subsequent weight can be a daunting endeavor, as there are over 740 DRG categories, with each DRG having one, two, or three subcategories based on comorbidities. In the evaluation of neonates, assigning a DRG is fairly simple as it is based on gestation, significant, and major comorbidities, or extreme immaturity or presence of RDS. In total, there are only seven separate DRGs to consider (Table 1). Determining presence of significant or major comorbidities is also fairly straightforward as demonstrated in Table 2. The corresponding weight carried by the neonatal DRG allows the clinician to assign a CMI for any population of similar patients, and using our graph compares a group along the spectrum of CMI values. In this study, we chose to evaluate the CMI for MM, MD, and DD twins born at our institution. As expected, the CMI of twins is higher than singletons born at our institution. For the neonatal population, DRGs and resulting CMI for a specific population can be used in functions aside from hospital billing. The CMI can be calculated for any neonatal population subset, and if plotted on the CMI graph (Figs. 1 and 2), a clinician can get a good idea of expected morbidities ranging from totally normal to extreme fetal immaturity at 22 weeks.

The in utero risks of twin pregnancies, including cord entanglement, twin–twin transfusion syndrome, and discordant growth, are well described based on amnionicity and chorionicity. Many investigators have described postnatal complications for twins based on chorionicity alone. Studies in the 1990s demonstrated that monochorionic twins (MM and MD combined) had increased rates of preterm birth, growth restriction, mortality, and the combined outcome of death or neurodevelopmental disability at 1 year of age.^12-15 Monochorionic twins also have higher rates of mortality, NEC, and cystic periventricular leukomalacia.¹⁶ A 2012 Danish cohort of over 2,000 diamniotic twin pregnancies (MD and DD combined) showed more fetal losses before 24 weeks’ gestation in MD pregnancies, but over 95% survival at 28 days if the fetus was live-born after 24 weeks’ gestation.⁷ Our study demonstrates that twin gestations in the current epoch of care, wherein the limits of viability have decreased to a lower GA are still at increased risk of morbidity and mortality during their entire hospital stay after birth, and MM twins in our cohort had the highest risk for postnatal complications. The increased morbidity in MM twins may be related to their earlier GA at birth, with all the associated comorbidities of extreme immaturity and/or intrauterine growth restriction.^7,13,15

Neonatologists are often asked to predict the postnatal course with limited information during prenatal counseling. DRGs, DRG weights, and resultant CMI have been used to predict survival and length of hospitalization among adult trauma patients¹⁷ and cost of hospitalization in adult inpatients with medical illness.¹⁸ Using individual patient DRG weight and resulting CMI of twin sets compared with the CMI of normal newborns, singleton admissions to the NICU, as well as data from the literature for term infants with significant congenital heart disease, and infants born at 22 weeks can be assessed. We are able to concisely summarize twin outcomes based on chorionicity and amnionicity.

Study limitations include the retrospective nature of this single-center study. Only complications documented in the problem list of the electronic medical record were factored into the DRG, and thus, some aspects of illness could be missed, though a single “data champion” was responsible for maintaining the problem lists through the duration of the evaluation period. Only complications that arose during the NICU course, and not long-term neurodevelopmental complications were evaluated. Long-term complications may factor into families’ antenatal decision-making and are not captured by the DRG weight. Some physicians and families may still prefer to have separate indices of mortality and morbidity, instead of the combined view offered by CMI for an entire population.

Our investigation highlights that twins and particularly MM twins experience increased morbidity and mortality compared with singletons, even those singletons admitted to the NICU. These differences in outcome are not captured by birth weight, GA, or hospital LOS alone. DRG weight, and CMI provide a readily available means of summarizing the higher mortality and complexity of the NICU course in twins. Because of its use in billing, DRG weight is already calculated for all hospitalized patients and thus can be used to create a summary of a single institution’s experience with any neonatal condition. However, with the relative simplicity of determining neonatal DRGs, clinicians can accurately assign an appropriate DRG usually within several days of birth. More research is needed to evaluate how a numeric summary such as DRG weight, and the CMI graph that we constructed can be utilized by neonatologists to evaluate clinical differences among similar and dissimilar patient groups. Knowledge of CMI for a specific neonatal population at the local level could help convey pertinent outcomes data to families during prenatal counseling.

Key Points.

• Using diagnosis-related group and case mix index to assess morbidities.

• Morbidities of twins are monochorionic-monoamniotic versus monochorionic-diamniotic versus dichorionic-diamniotic twins.

• Only seven diagnosis-related group in neonatology make it a valuable tool for clinicians.