Page 1
RESEARCH ARTICLE Open Access
Evaluation of hospital inpatient complications:
a planning approach
Ronald J Lagoe1, Gert P Westert2*
Abstract
Background: Hospital inpatient complications are one of a number of adverse health care outcomes. Reducing
complications has been identified as an approach to improving care and saving resources as part of the health
care reform efforts in the United States.
An objective of this study was to describe the Potentially Preventable Complications software developed as a tool
for evaluating hospital inpatient outcomes. Additional objectives included demonstration of the use of this soft-
ware to evaluate the connection between health care outcomes and expenses in United States administrative data
at the state and local levels and the use of the software to plan and implement interventions to reduce hospital
complications in one U.S. metropolitan area.
Methods: The study described the Potentially Preventable Complications software as a tool for evaluating these
inpatient hospital outcomes. Through administrative hospital charge data from California and Maryland and
through cost data from three hospitals in Syracuse, New York, expenses for patients with and without
complications were compared. These comparisons were based on patients in the same All Patients Refined
Diagnosis Related Groups and severity of illness categories. This analysis included tests of statistical significance.
In addition, the study included a planning process for use of the Potentially Preventable Complications software in
three Syracuse hospitals to plan and implement reductions in hospital inpatient complications. The use of the PPC
software in cost comparisons and reduction of complications included tests of statistical significance.
Results: The study demonstrated that Potentially Preventable Complications were associated with significantly
increased cost in administrative data from the United States in California and Maryland and in actual cost data
from the hospitals of Syracuse, New York. The implementation of interventions in the Syracuse hospitals was
associated with the reduction of complications for urinary tract infection, decubitus ulcer, and pulmonary
embolism.
Conclusions: The study demonstrated that the Potentially Preventable Complications software could be used to
evaluate hospital outcomes and related costs at the aggregate and diagnosis specific levels. It also indicated that
the system could be used to plan and implement interventions to improve outcomes on an individual or
multihospital basis.
Background
Interest in the improvement of quality and outcomes in
health care is increasing in both Europe and the United
States. Initially, this development was driven mainly by
attention to improving care at the patient level [1].
More recently, however, a linkage between improving
outcomes and saving health care costs has developed.
An important study evaluated the direct medical costs
of a wide range of adverse events in hospitals in the
Netherlands. This study included unplanned admissions;
unplanned readmissions; hospital acquired complica-
tions including infections, neurological complications,
and other diagnoses; unexpected mortality, and other
issues. A study of a wide range of adverse medical
events in hospitals in the Netherlands suggested that
30,000 inpatient admissions included a preventable
adverse event. On the basis of data collected in 2004, it
estimated that the annual costs of preventable adverse
* Correspondence: gert.westert@rivm.nl
2National Institute for Public Health and the Environment Bilthoven,
Netherlands
Lagoe and Westert BMC Health Services Research 2010, 10:200
http://www.biomedcentral.com/1472-6963/10/200
© 2010 Lagoe and Westert; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative
Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and
reproduction in any medium, provided the original work is properly cited.
Page 2
events in Dutch hospitals was 161 million euros [2].
Other European studies have evaluated potentially pre-
ventable mortality in Dutch hospitals [3], hospital
acquired urinary tract infections [4], nosocomial pneu-
monia [5,6], hospital acquired infections [7], pressure
ulcers [8], and adverse events in British Hospitals [9].
The Hoonhout and Zegers studies were carried out
through the Netherlands adverse events research and
used the same databases.
Studies have also evaluated the financial impact of
adverse health care outcomes and related issues in the
United States including surgical site infections [10,11],
urinary tract infections [12], clostridium difficile [13],
and hospital acquired pressure ulcers [14]. Like their
European counterparts, some of these have included
efforts to address the relationship between health care
outcomes and costs [15].
The context of health care reform in national politics
has become increasingly important for evaluation and
improvement of hospital outcomes. This context relates to
the process of health care reform and the development of
software to define and analyze these outcomes. Despite
increasing interest in the subject, research concerning the
costs of adverse health care outcomes and efforts to
address them have been hampered by the lack of major
provider incentives for quality improvement and the diffi-
culties of data collection in this area. Recent developments
suggest that situation is changing in the United States.
Evidence is growing that the health care reform move-
ment in the United States that was stimulated by the
Obama Administration will attempt to reduce provider
reimbursement for adverse clinical outcomes. A major
driver of this process has been increased public aware-
ness of the high level of health care spending in the
United States. Within the health care reform debate, the
Obama Administration has repeatedly emphasized
the lack of connection between high levels of health
care expenditures and outcomes such as life expectancy
and mortality compared with other nations [16-18].
These issues have been complicated by the impact of
the economic recession of 2008 and 2009. The need for
increased government spending by the U.S. federal gov-
ernment to stimulate economic recovery has been con-
strained by the continued increase in public expenditures
for health care through the Medicare and Medicaid pro-
grams [19]. In his Inaugural Address, President Obama
called for higher quality of care at reduced costs [20].
In this context, it has been suggested that inpatient
hospital complications generate large expenses in Eur-
ope and the United States. These complications are
diagnoses which developed after admission to inpatient
hospital beds. These complications are defined as harm-
ful events or negative outcomes occurring during inpati-
ent hospitalization that result from the processes of care
and treatment rather than the natural progression of
diseases [11].
Inpatient hospital complications are important because
they can result in substantial adverse outcomes for
patients. Events such as infections and circulatory com-
plications cause considerable discomfort for patients,
sometimes with life threatening consequences. These
events disrupt the patients’ recovery as well as their
return to residential environments and employment.
Because complications can extend hospital stays and
consume additional resources, they can also block access
to hospital beds for other patients.
Inpatient complications are also costly. They result
in large expenditures for additional nursing time,
pharmaceuticals, and tests before discharge is possible.
Although the frequency of complications is often low,
the total additional expenditures can be large at the pro-
vider and system wide levels [16].
Fortunately, the rising interest in improving the quality
of care as a means of reducing costs are being supported
by the availability of new resources for the collection and
analysis of health care outcomes data. These include new
administrative data categories which address outcomes
such as inpatient hospital complications and computer
software such as the Potentially Preventable Complica-
tions system developed by the 3M™ Corporation which
was created to analyze these indicators [21]. The Poten-
tially Preventable Complications system was developed to
evaluate inpatient hospital outcomes based on adminis-
trative data. The system has been validated in preliminary
studies by the New York State Department of Health and
the Massachusetts Hospital Association. The system is
currently being evaluated in a series of demonstration
projects in the United States.
This objective of this study was to describe the Poten-
tially Preventable Complications software as a tool for
evaluating hospital inpatient outcomes. The study
showed the use of this software to analyze the connec-
tion between health care outcomes and expenses in Uni-
ted States administrative data at the state and local
levels. The study also described the use of the software
to plan and implement interventions to reduce hospital
complications in one U.S. metropolitan area.
The data used in this study are openly available.
Administrative databases include patient specific data
but may, in some cases, limit the availability of unique
patient identifiers.
Methods
Potentially Preventable Complications
Potentially Preventable Complications is a system for
categorizing and evaluating inpatient hospital complica-
tions developed by the 3M™ Corporation. The following
material describes the specific components of this
Lagoe and Westert BMC Health Services Research 2010, 10:200
http://www.biomedcentral.com/1472-6963/10/200
Page 2 of 12
Page 3
system within the context of hospital outcomes
improvement.
The PPC system provides information concerning a
full range of complications, as well as individual cate-
gories for health care providers [21]. The complications
included in the system were identified through a review
of existing literature, the diagnosis codes used in the
Complications Screening Protocol developed by Iezzoni
et al, and the Patient Safety Indicators developed by the
Agency for Healthcare Research and Quality [22,23].
The full list of PPCs includes 64 mutually exclusive
categories that were identified from ICD-9-CM proce-
dure codes for secondary diagnoses. These diagnoses are
identified as clinical conditions that were not the princi-
pal causes of hospital admission. The 64 categories have
been condensed to a summary list of 35 PPCs. This pro-
cess was carried out in order to support clinical man-
agement of related diagnoses in hospitals. It was based
on clinical management needs, rather than scientific evi-
dence concerning these diagnoses. 3M™ Health Informa-
tion Services is currently developing a revised PPC
algorithm based on ICD-10 codes. That process is
scheduled for completion in 2011. The summary list fol-
lows in Table 1.
Under the Potentially Preventable Complication sys-
tem, a number of diagnoses are excluded as non preven-
table. These include complications directly related to
malignant diseases, multiple trauma, organ transplants,
specific burns, and HIV related disorders. Because of
their unique characteristics, neonates are also excluded.
A number of these diagnoses are excluded by the algo-
rithm as global exclusions, whether they appear as prin-
cipal or secondary diagnoses. Others are excluded only
when accompanied by another specific diagnosis [21].
The Potentially Preventable Complications system
employs administrative hospital data. Within these
data, the Present on Admission indicator is used to
identify diagnoses which develop after admission to
acute hospitals. The Present on Admission Indicator is
applied to each secondary diagnosis of each hospital
inpatient. Each of these diagnoses are identified as
present on admission, not present on admission, or
undetermined.
Use of the Present on Admission indicator on a wide-
spread basis in the United States was made possible by
National Uniform Billing Committee changes (UB 04)
approved May 23, 2007. Under these changes, the stan-
dard inpatient claim form was modified to allow the
submission of a Present on Admission indicator for
each secondary diagnosis [24]. A number of state data
bases have followed Medicare by requiring hospitals to
use this indicator.
The validity of the Present on Admission indicator is
based on the degree of compliance of hospital staffs
with Medicare criteria for this indicator. It was man-
dated for use throughout the United States by Medicare
in 2007. In New York State, this indicator has been used
since the mid 1990s. It is assumed that the extensive
experience of New York State and the Syracuse hospitals
with the Present on Admission indicator contributed to
the validity of related data used to define PPCs [25]. At
the same time, this experience is based on unpublished
data which has not benefited from extensive evaluation
Table 1 PPC Summary Categories
01 Stroke & Intracranial Hemorrhage
02 Extreme CNS Complications
03 Acute Pulmonary Edema and Respiratory Failure with Mechanical
Ventilation
04 Pneumonia & Other Lung Infections
05 Aspiration Pneumonia
06 Pulmonary Embolism
07 Shock
08 Congestive Heart Failure
09 Acute Myocardial Infarct
10 Ventricular Fibrillation/Cardiac Arrest
11 Peripheral Vascular Complications Except Venous Thrombosis
12 Venous Thrombosis
13 Major Gastrointestinal Complications with Transfusion or Significant
Bleeding
14 Major Liver Complications
15 Clostridium Difficile Colitis
16 Urinary Tract Infection
17 Renal Failure with Dialysis
18 Post-Hemorrhage & Other Acute Anemia with Transfusion
19 Decubitus Ulcer
20 Septicemia & Severe Infections
21 Post-Op Wound Infection & Deep Wound Disruption with
Procedure
22 Reopening Surgical Site
23 Post-Op Hemorrhage & Hematoma with Hem Cntrl Proc or I&D
Proc
24 Accidental Puncture/Laceration during Invasive Procedure
25 Post-Procedure Foreign Bodies
26 Encephalopathy
27 Iatrogenic Pneumothrax
28 Mechanical Complication of Device, Implant & Graft
29 Inflammation & Other Complications of Devices, Implants or Grafts
Except Vascular Infection
30 Infections Due to Central Venous Catheters
31 Obstetrical Hemorrhage with Transfusion
32 Obstetric Laceration & Other Trauma without Instrumentation
33 Obstetric Laceration & Other Trauma with Instrumentation
34 Major Puerperal Infection and Other Major Obstetric Complications
35 Post-Op Respiratory Failure with Tracheostomy
Lagoe and Westert BMC Health Services Research 2010, 10:200
http://www.biomedcentral.com/1472-6963/10/200
Page 3 of 12
Page 4
and scrutiny. For this reason, additional evaluation of
the reliability of the Present on Admission data is
necessary.
Under the Potentially Preventable Complications sys-
tem, candidate diagnoses are those that are not excluded
at the global or specific levels. These diagnoses are iden-
tified as not present on hospital admission. The candi-
date diagnoses are evaluated with respect to risk of
complications based on reasons for admission and sever-
ity of illness. Evaluation of risk of complications based
on reason for admission is related to hospital services
provided to each patient. A major factor in this evalua-
tion is the type of hospital service, whether medical or
surgical, that a patient receives. The susceptibility of
complications varies widely among medical and surgical
patients. For example, among medicine patients, aspira-
tion pneumonia is more likely to develop as a complica-
tion for a patient with stroke as a principal diagnosis
than for one with urinary retention. The type of surgery
is also a major influence on the likelihood of specific
complications [21].
Risk of complications is also closely related to the
severity of illness of each patient. Those hospitalized with
severe secondary diagnoses or with multiple comorbid-
ities have a higher risk of developing complications than
those who do not [26,21]. The Potentially Preventable
Complication system employs the All Patients Refined
algorithm to determine severity of illness. This algorithm
employs degree of illness based on principal and second-
ary diagnoses identified by (ICD-9-CM) medical record
codes, patient age, and other factors [27].
The use of severity of illness to evaluate risk of hospi-
tal complications makes it possible to compare risk
levels within or between hospitals under the Potentially
Preventable Complications System. Within this context,
comparisons can be developed within the same hospital,
for different hospitals, or between hospitals and regional
benchmark populations. For this reason, comparisons of
populations need to be stratified by Diagnosis Related
Group and severity of illness. This capability is limited
by the fact that secondary diagnoses and comorbidities
that are complications also influence severity of illness
within Diagnosis Related Groups. For example, the
development of pneumonia as a post admission compli-
cation could cause the severity of illness of a medical or
surgical inpatient to increase from 1 (Minor) to 2 (Mod-
erate). A post admission complication for urinary tract
infection would have a similar impact. The impact of
this phenomenon can be limited by the fact that inpati-
ent complication rates are generally relatively low.
The Potentially Preventable Complications system
has made it possible to evaluate the financial impact
of these outcomes in acute hospitals. Because compli-
cations are identified by All Patients Refined Severity of
Illness, it is possible to identify at risk populations and
populations who experience each complication by DRG
and severity. The hospital charges for these populations
can then be compared using administrative data. These
data are based on reported charges by hospital. Because
charges are all inclusive, they can contain actual costs as
well as provider mark ups. In the study cited, they were
assumed to be generally representative of actual hospital
costs. It must be emphasized, however, that the quanti-
tative relationship varies by hospital because of differ-
ences in cost accounting and provider mark ups [24].
The Potentially Preventable Complications system is
currently being tested at a number of hospitals in the
United States. State governments and private organiza-
tions will be using the system for public and provider
reporting of outcomes data.
Potentially Preventable Complications Applied
The application of the Potentially Preventable Complica-
tions system to hospital quality improvement is illu-
strated by the experiences of the acute care facilities of
Syracuse, New York. The metropolitan area of Syracuse,
New York includes a resident population of 446,065
[28] and four urban hospitals. The hospitals, which gen-
erate approximately 70,000 discharges annually, have
historically worked to improve efficiency and outcomes
through their cooperative planning organization, the
Hospital Executive Council [29].
The Syracuse hospitals became interested in using the
Potentially Preventable Complications system for differ-
ent reasons. One of these was the commitment of the
Obama Administration to reducing health care costs by
improving the quality of care. The Administration has
suggested that reduction of complications will be an
important part of its health care reform program. The
Obama Administration has indicated that it will be
developing financial mechanisms for reducing complica-
tions as part of the implementation of its health care
reform program during 2011 and 2012. In New York
State, reduction of complications through the use of
PPCs has been included in the Governor’s proposed
budget for 2010 - 2011.
Another major influence on the Syracuse hospitals was
the potential for internal cost savings. Preliminary inpa-
tient data developed by the Hospital Executive Council
and the Syracuse hospitals suggested that the actual
costs of treating patients with PPCs was substantially
higher than the costs of treating patients with the same
APR DRGs and severity levels without the complica-
tions. The data were based on actual hospital costs
rather than charges. These costs were developed using
cost accounting systems at the individual hospitals.
In this study, the Potentially Preventable Complica-
tions categories were employed as a component of the
Lagoe and Westert BMC Health Services Research 2010, 10:200
http://www.biomedcentral.com/1472-6963/10/200
Page 4 of 12
Page 5
Method. This was based on the need to test the use of
these categories in addressing inpatient complications. It
was not within the scope of this study to test or adjust
the content of the PPC categories.
The program for reduction of Potentially Preventable
Complications in the Syracuse hospitals was developed
jointly by hospital staff members identified by the Chief
Executive Officers, the Hospital Executive Council staff,
and representatives of 3M™ Health Information Services.
It included the following components.
Data Development and Distribution
Educational Program
Clinical Record Reviews
Selection of Objectives by PPC Category
Identification of Drivers of Complications
Identification and Implementation of Interventions
At the beginning of the project, the 3M™ Potentially
Preventable Complications Software was still under devel-
opment and could not be transferred. As a result, the
initial monitoring data were produced directly by 3M™
and distributed in Syracuse through the Hospital Execu-
tive Council. These data included summary tables con-
taining PPC frequencies and rates per at risk discharges
for the individual hospitals and severity adjusted bench-
mark populations. These data were produced for the 35
PPC summary categories. They also included patient
specific summary data for the hospital PPC and at risk
populations. Initial data provided were for the periods
January - December 2007 and January - June 2008.
The project also involved presentation of an educa-
tional program to key hospital staff members. The
program involved teleconferences concerning the Poten-
tially Preventable Complications System by senior staff
of 3M™ Health Information Services to hospital adminis-
trators, physicians, and quality assurance staff. These
presentations described the development and structure
of the PPC algorithm, as well as its potential uses to
improve outcomes and reduce costs.
Following the educational presentations, each hospital
used the Potentially Preventable Complications informa-
tion which had been provided to conduct reviews com-
paring the information produced by the software with
clinical data in patient charts. At three of the hospitals,
the data produced by the software was found to be con-
sistent with data in the charts. At the fourth hospital, it
was determined that the existence of numerous false
positives in the complications data would require a
recoding process. This process was necessary to verify
the presence of inpatient complications based on docu-
mentation in the patient records.
The next phase of the process was identification of
Potentially Preventable Complications categories that
were to be the objectives of the quality assurance
process. Within each of these categories, the hospitals
identified drivers of the complications and developed
interventions to address the drivers. These activities
were followed by the implementation of interventions
between October 2008 and March 2009.
The interventions developed in the Syracuse hospitals
focused on urinary tract infections, pneumonia, clostri-
dium difficile colitis, decubitus ulcer, pulmonary embo-
lism, and post hemorrhagic and other acute anemia with
transfusion. Each hospital identified specific PPCs that
would be the focus of its interventions. These selections
were based on historical complication rates, hospital
resources, and quality assurance objectives. The data in
Table 2 identify numbers of hospital complications and
complication rates compared with severity adjusted
benchmarks for each PPC that was a focus on an inter-
vention by the participating hospitals.
Interventions addressing urinary tract infections at
Community-General and Crouse hospitals focused on
development of protocols for urine sample collection
and nursing procedures regarding catheter insertion,
removal, and care. Educational programs were devel-
oped to implement these procedures at the two hospi-
tals. At Crouse Hospital, stickers were developed for
attachment to patient charts to remind nurses of the
need for vigilance in this area. At Community-General,
stamps were placed in patient charts to remind physi-
cians to reorder or discontinue the catheters. At both
hospitals, interventions also included daily monitoring
of patients for sites of infection and laboratory work.
The Crouse and Community-General programs were
implemented in October 2008.
For pneumonia at St. Joseph’s Hospital Health Center,
a large majority of the complications were produced by
surgery, especially cardiac and orthopedic surgery. It
was determined that the principal drivers of this com-
plication were ventilator use and failure to implement
early ambulation of post surgery patients. In order to
address this complication, the hospital developed inter-
ventions addressing ventilator associated pneumonia
in October 2008. Additional protocols addressing
community-acquired pneumonia, which included expe-
dited ambulation of patients after surgery, were also
implemented.
For clostridium difficile colitis (c. diff.) at Community-
General Hospital, the principal driver was determined to
be environmental because this infection can remain in
the environment for long periods of time and cultured
from almost any surface. Complications of influenza
were also a potential driver, since they disrupt normal
intestinal flora, leading to an overgrowth of c diff. Inter-
ventions were identified for this PPC including
improved cleaning procedures in treatment and patient
Lagoe and Westert BMC Health Services Research 2010, 10:200
http://www.biomedcentral.com/1472-6963/10/200
Page 5 of 12
End of preview.
