BACKGROUND:

As payment models continue to move toward value-driven care, the quality of documentation has become more important than ever. Clinical Documentation Integrity (CDI) programs can aid in the documentation of diagnoses that are specific and consistent throughout the medical record, which leads to accurate code assignment, better understanding of patient complexity, and improved facility reimbursement.

METHODS:

An interrupted time series analysis was conducted by using a segmented regression model to estimate the impact of our hospital’s CDI program on perceived patient complexity using severity of illness stratification, observed to expected mortality ratio and case-mix index. Patients who died during the admission were chosen to limit our analysis to patients with the highest severity of illness.

RESULTS:

A total of 206 patients who had died while inpatient at our 400 bed children’s hospital were included. There was a 15.7% increase in patients who were final coded with the highest level of severity of illness after our CDI program launched compared with those patients admitted before program inception. The hospital case-mix index for inpatient cases increased 25% from 2011 to 2017. There was a 44% decrease in the observed to expected mortality ratio.

DISCUSSION:

A CDI program can have a significant impact, as evidenced by our ability to show complexity gains on some of the sickest patients by supporting documentation of precise, accurate diagnoses. In turn, this may allow for better understanding of the complexity of our patient population and support appropriate reimbursement and payer contract negotiations.

As health care continues to move toward a more value-based model,13  it is imperative for providers to accurately demonstrate the complexity of their patient population.4  Metrics such as case-mix index (CMI) and hospital-specific mortality rates are being reported publicly, and it is increasingly important for providers to better understand how to stratify their patient populations. This enables health care providers to tailor strategies appropriate to each patient and direct resources toward interventions that will make the biggest impact. As an example, one children’s hospital used comparative data within a large, national pediatric database to identify a severity-adjusted length of stay and cost outlier population (patients with low-acuity asthma). A quality improvement initiative was undertaken, and the hospital reduced both length of stay and decreased cost for this cohort, in addition to optimizing diagnosis documentation to accurately reflect patient acuity.5  A common way to stratify pediatric inpatients is by All Patient Refined Diagnosis Related Group (APR-DRG) metrics, used by both Medicaid (the single largest insurer of children) and commercial insurance companies.6,7  The APR-DRG system assigns patients 1 of 4 risk of mortality subclasses and 1 of 4 severity of illness (SOI) subclasses: minor (1), moderate (2), major (3), and extreme (4).8  CMI represents the average SOI of all patients in a hospital on the basis of the diagnosis related group (DRG) in which each patient falls. Because the APR-DRG methodology relies on complete documentation of pertinent diagnoses in the medical record, inadequate documentation may lead to an inaccurate reflection of patient complexity. Clinical Documentation Integrity (CDI) programs review medical records and seek clarification in the form of queries when documentation of diagnoses is incomplete, imprecise, or inconsistent. Accurate and complete documentation of specific diagnoses allows hospital coders to correctly assign principal and secondary diagnoses, which in turn reflects the severity of patient illness.

In addition to individual patient metrics such as SOI, another key area for CDI involvement lies within quality reporting.9  Ensuring documentation supports the level of care provided within a hospital stay has ramifications for how hospitals and physicians are viewed in this area.10,11  Additionally, many of these reports can be found publicly, which can influence patient choice in providers. One way that quality is measured is through mortality rates based on observed to expected deaths (O/E ratio). An O/E ratio of <1 indicates that the mortality rate at a specific hospital is less than expected on the basis of that institution’s case mix, and the opposite is true for an O/E ratio of >1. Severity-adjusted O/E ratios are used to compare the performances of hospitals with the goal of identifying potential quality improvement opportunities.12,13  One method to calculate the expected number of deaths is to identify all patients within a cohort who had a risk of mortality score of 4 based on the APR-DRG system, because this has been found to correlate closely with mortality.14  We sought to determine if the perceived complexity of this extremely ill patient population increased after the inception of our CDI Program. Specifically, we hypothesized that improved documentation of clinically relevant diagnoses would lead to an appropriate increase in the number of patients with a final coded SOI of 4, an increase in CMI, and a decrease in the O/E ratio.

In 2014, our 400-bed freestanding children’s hospital introduced a CDI program with the intent of demonstrating the high-quality care provided to patients by ensuring clear and precise clinical documentation. During the initial phases of program development, a steering committee was formed with representation from hospital leadership, physicians, Case Management, Health Information Management, the Compliance Department, and Finance. Additionally, we engaged a consulting firm to guide the implementation process. In conjunction with a process to query prescribers for unclear or missing diagnoses, our program provided clinician education, distributed documentation tips, and communicated with departmental physician champions.

Data were analyzed from 2011 to 2017 with the following criteria: patients with a discharge disposition of death and who were admitted to an inpatient level of care from 2011 to 2017 with specifically identified payers, 2 commercial and 1 government. By choosing patients who died during the admission, we sought to limit our analysis to the highest SOI. We limited our analysis to these specific payers because they contracted with our hospital based on a DRG reimbursement methodology. We excluded patients from 2014 because this year corresponded with the inception of our CDI Program. Specifically, our program launched in June 2014 and required a ramp-up period of ∼6 months because of the hiring and training of new CDI specialists as well as optimizing workflow (transitioning from e-mail to electronic medical record queries). Therefore, we excluded the 6 months before and after launch of the program from the analysis. We divided cases into pre-CDI (2011–2013) or post-CDI (2015–2017). Using a commercially available software product, we converted all patients into APR-DRG version 30 (3M APR DRG Software, St Paul, MN) for an equitable comparison.

Interrupted time series analysis was conducted by using a segmented regression model (an impact model of level change) to estimate the impact of the CDI intervention on increased SOI 4 coding, taking into account baseline trends. We also compared CMI as well as the O/E ratio before and after CDI program inception. All analyses were performed with SAS version 9.4 (SAS Institute, Cary, NC).

Two hundred and six patients met criteria for inclusion. There were no significant differences in the patient population before and after CDI program inception for service line designation, length of stay, and age. More patients in the post-CDI group were insured by the government (P = .023; pre-CDI = 51%, post-CDI = 67%) (Table 1). The number of deaths within the identified period was relatively consistent (P = .51; 2011 = 29, 2012 = 40, 2013 = 40, 2015 = 30, 2016 = 36, 2017 = 31). There was a 15.7% increase in patients final coded with an SOI of 4 during the 2015–2017 time period compared with those patients admitted in 2011–2013 (P = .02) (Fig 1). The hospital CMI for inpatient cases increased 24.6% from 2011 to 2017 (P = <0.001; 2011 = 1.63, 2017 = 2.03). Within the pre-CDI period, there were 109 observed deaths, whereas 269 were expected (O/E ratio = 0.41). For the post-CDI period, there were 97 deaths, whereas 431 were expected (O/E ratio = 0.23) (Fig 2).

TABLE 1

Patient Characteristics

CharacteristicPre-CDI Program (n = 109)Post-CDI Program (n = 97)P
Service line, n (%)   .069 
 Critical care 33 (30) 35 (36)  
 Cardiovascular 18 (17) 19 (20)  
 General pediatrics 45 (41) 22 (23)  
 Oncology 8 (7) 12 (12)  
 Surgery 4 (4) 5 (5)  
 Neurology 1 (1) 4 (4)  
Length of stay, n (%), d   .889 
 1–3 34 (31) 26 (27)  
 4–7 12 (11) 13 (13)  
 8–14 18 (17) 16 (16)  
 ≥15 45 (41) 42 (43)  
Age, n (%)   .236 
 0–28 d 33 (30) 21 (22)  
 29 d to 2 y 18 (17) 20 (21)  
 >2 and <5 y 5 (5) 10 (10)  
 ≥5 y 53 (49) 46 (47)  
Payer, n (%)   .023 
 Government 56 (51) 65 (67)  
 Commercial 53 (49) 32 (33)  
CharacteristicPre-CDI Program (n = 109)Post-CDI Program (n = 97)P
Service line, n (%)   .069 
 Critical care 33 (30) 35 (36)  
 Cardiovascular 18 (17) 19 (20)  
 General pediatrics 45 (41) 22 (23)  
 Oncology 8 (7) 12 (12)  
 Surgery 4 (4) 5 (5)  
 Neurology 1 (1) 4 (4)  
Length of stay, n (%), d   .889 
 1–3 34 (31) 26 (27)  
 4–7 12 (11) 13 (13)  
 8–14 18 (17) 16 (16)  
 ≥15 45 (41) 42 (43)  
Age, n (%)   .236 
 0–28 d 33 (30) 21 (22)  
 29 d to 2 y 18 (17) 20 (21)  
 >2 and <5 y 5 (5) 10 (10)  
 ≥5 y 53 (49) 46 (47)  
Payer, n (%)   .023 
 Government 56 (51) 65 (67)  
 Commercial 53 (49) 32 (33)  
FIGURE 1

Time series analysis demonstrating the number of patients final coded with an SOI of 4 before and after the intervention. CY, calendar year; Q, quarter.

FIGURE 1

Time series analysis demonstrating the number of patients final coded with an SOI of 4 before and after the intervention. CY, calendar year; Q, quarter.

Close modal
FIGURE 2

O/E mortality ratio pre- and post-CDI period.

FIGURE 2

O/E mortality ratio pre- and post-CDI period.

Close modal

The importance of accurate, consistent, and specific documentation of diagnoses is better known in adult medicine compared with pediatrics because of the interaction of documentation with reimbursement and quality ratings in the Medicare population as well as the fact that pediatrics is relatively new to CDI.4,15  However, payment models in pediatrics continue to move toward value-driven care. Consequently, the quality of clinical documentation has become more important than ever, and pediatricians may not be well versed in this subject matter. Moreover, we were unable to find any data in the literature describing the return on investment for pediatric CDI programs, which may be helpful information for institutions considering developing such an initiative.

We found that a pediatric CDI program can have a significant impact, as evidenced by our ability to demonstrate complexity gains in the sickest patients. In this study, we demonstrated an increase in patients final-coded with a SOI of 4, a decreased O/E mortality ratio, and an increase in hospital CMI after the inception of a CDI program. The most-plausible interpretation of our data is that the pre- and post-CDI populations were similarly complex but the SOI of the pre-CDI cohort was not being accurately reflected in the coded data. Our evidence supports that these results are not due to a change in the base population but rather to enhanced documentation within the cohort to more accurately reflect the complexity of our patient population. In fact, the notable increase in number of patients with SOI of 4 during quarter 3 of 2014 corresponded with the launch of our CDI program.

In addition to effects on quality reporting, increasing CMI and SOI can also have financial implications. CMI can affect payment contracts in bundled and capitation models. Furthermore, DRGs are assigned a numeric value, the relative weight, which reflects how acutely ill the patients are in that particular DRG. Therefore, relative weight is one of the major determinants of reimbursement for health care institutions in this model, and it is affected by the quality of documentation of specific diagnoses.16 

We recognize there are limitations to our data, most specifically the comparison of the pre-CDI and post-CDI program groups. There were more patients who were insured by the government in the post-CDI cohort, which may indicate more negative social determinants of health. In addition, there could have been shifts in our patient population or differences in the surgical procedures performed. However, we found that both the pre-CDI and post-CDI patient populations were comparable from a service line perspective, which suggests large shifts did not occur. Furthermore, the yearly number of deaths during the study period did not differ significantly.

Pediatric CDI program review of charts can have an impact on final code assignment by supporting documentation of precise, accurate diagnoses. In turn, this may allow for better understanding of the complexity of our pediatric patient population and support appropriate reimbursement and payer contract negotiations.

Ms Reardon conceptualized and designed the study, collected data, and contributed to drafting the initial manuscript; Ms Foley conducted analyses and contributed to drafting the initial manuscript; Ms Melvin conducted analyses; Dr Agus participated in study conceptualization and design; Dr Sanderson participated in study conceptualization and design, collected data, and contributed to drafting the initial manuscript; and all authors critically reviewed and revised the manuscript and approved the final manuscript as submitted.

FUNDING: No external funding.

1
Centers for Medicare & Medicaid Services
.
2
Saleh
KJ
,
Shaffer
WO
.
Understanding value-based reimbursement models and trends in orthopaedic health policy: an introduction to the Medicare Access and CHIP Reauthorization Act (MACRA) of 2015
.
J Am Acad Orthop Surg
.
2016
;
24
(
11
):
e136
e147
3
Rajaram
R
,
Barnard
C
,
Bilimoria
KY
.
Concerns about using the patient safety indicator-90 composite in pay-for-performance programs
.
JAMA
.
2015
;
313
(
9
):
897
898
4
Kulkarni
D
,
Heath
J
,
Kosack
A
,
Jackson
NJ
,
Crummey
A
.
An educational intervention to improve inpatient documentation of high-risk diagnoses by pediatric residents
.
Hosp Pediatr
.
2018
;
8
(
7
):
430
435
5
Sedman
AB
,
Bahl
V
,
Bunting
E
, et al
.
Clinical redesign using all patient refined diagnosis related groups
.
Pediatrics
.
2004
;
114
(
4
):
965
969
6
Children’s Hospital Association
.
7
Maeda
J
,
Nelson
L
.
An analysis of private-sector prices for hospital admissions. 2017. Available at: https://www.cbo.gov/system/files/115th-congress-2017-2018/workingpaper/52567-hospitalprices.pdf. Accessed August 4, 2020
8
Averill
RF
,
Goldfield
N
,
Hughes
J
, et al
.
3M APR DRG classification system version 26.1. Methodology overview. 2008. Available at: www.hcup-us.ahrq.gov/db/nation/nis/v261_aprdrg_meth_ovrview.pdf. Accessed August 4, 2020
9
Loftus
T
,
Najafian
H
,
Pandey
SR
,
Ramanujam
P
.
The impact of documentation training on performance reporting
.
Cureus
.
2015
;
7
(
7
):
e283
10
McDonald
KM
,
Davies
SM
,
Haberland
CA
,
Geppert
JJ
,
Ku
A
,
Romano
PS
.
Preliminary assessment of pediatric health care quality and patient safety in the United States using readily available administrative data
.
Pediatrics
.
2008
;
122
(
2
).
11
Short
HL
,
Heiss
KF
,
Wulkan
ML
,
Raval
MV
.
Clinical validity and relevance of accidental puncture or laceration as a patient safety indicator for children
.
J Pediatr Surg
.
2017
;
52
(
1
):
172
176
12
Wiebe
DJ
,
Holena
DN
,
Delgado
MK
,
McWilliams
N
,
Altenburg
J
,
Carr
BG
.
The Pennsylvania trauma outcomes study risk-adjusted mortality model: results of a statewide benchmarking program
.
Am Surg
.
2017
;
83
(
5
):
445
452
13
Gaies
M
,
Ghanayem
NS
,
Alten
JA
, et al
.
Variation in adjusted mortality for medical admissions to pediatric cardiac ICUs
.
Pediatr Crit Care Med
.
2019
;
20
(
2
):
143
148
14
Baram
D
,
Daroowalla
F
,
Garcia
R
, et al
.
Use of the All Patient Refined-Diagnosis Related Group (APR-DRG) risk of mortality score as a severity adjustor in the medical ICU
.
Clin Med Circ Respirat Pulm Med
.
2008
;
2
:
19
25
15
Association of Clinical Documentation Integrity Specialists
.
A new frontier: finding profiles in pediatric CDI. 2017. Available at: https://acdis.org/articles/new-frontier-finding-profiles-pediatric-cdi. Accessed August 4, 2020
16
Centers for Medicare & Medicaid Services
.
Acute inpatient PPS. Available at: https://www.cms.gov/medicare/medicare-fee-for-service-payment/AcuteInpatientPPS/index.html. Accessed August 4, 2020

Competing Interests

POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.

FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.