Evidence on Prognosis

Article (PDF Available)inPlastic and Reconstructive Surgery 131(3):655-65 · March 2013with32 Reads
DOI: 10.1097/PRS.0b013e31827c6f90 · Source: PubMed
Abstract
In the clinical decision-making process, prognostic information gives physicians guidance regarding disease and treatment that can be passed along to patients. Using the clinical scenario of a type IIIB open tibial fracture as an example, this article focuses on how to search prognostic literature effectively and how to critically appraise prognostic studies. A well-built clinical question using the Patients, Intervention, Comparison, and Outcomes methodology helps plastic surgeons derive information from the vast resources in an effective and time-efficient manner. Subsequent critical appraisal of the collected studies is the essential step for identifying the most relevant high-quality evidence to provide the best prognostic information to guide patient decision-making.

Figures

Figure
Figure

Full-text (PDF)

Available from: Sandeep Sebastin
EBM SPECIAL TOPIC
Evidence on Prognosis
Shimpei Ono, M.D., Ph.D.
Sandeep J. Sebastin,
M.Ch.(Plast.)
Kevin C. Chung, M.D., M.S.
Tokyo, Japan; Singapore; and
Ann Arbor, Mich.
Summary: In the clinical decision-making process, prognostic information gives
physicians guidance regarding disease and treatment that can be passed along
to patients. Using the clinical scenario of a type IIIB open tibial fracture as an
example, this article focuses on how to search prognostic literature effectively and
how to critically appraise prognostic studies. A well-built clinical question using the
Patients, Intervention, Comparison, and Outcomes methodology helps plastic sur-
geons derive information from the vast resources in an effective and time-efficient
manner. Subsequent critical appraisal of the collected studies is the essential step
for identifying the most relevant high-quality evidence to provide the best prog-
nostic information to guide patient decision-making. (Plast. Reconstr. Surg. 131:
655, 2013.)
The word “prognosis” is derived from the
Greek prognostikos. It combines pro (before)
and gnosis (knowledge). Hippocrates used
the word prognosis much as we do today, to pro-
vide important clinical information regarding pos-
sible outcomes (favorable and unfavorable) after
disease or treatment.1In the surgical field, prog-
nosis may be defined as predicting outcomes of a
surgical intervention based on current best evi-
dence. Prognostic information arms a surgeon
with data to guide patients in selecting the best
treatment based on the injury pattern and the
patient’s specific demographic characteristics and
needs.
The desired goal of discussing prognosis is
made more challenging by the relative lack of
these types of data in the plastic surgery literature.
To achieve the best outcome for each patient, it is
necessary for the plastic surgeon to become an
expert in methodically estimating the probabili-
ties of various events that affect outcomes occur-
ring along each available therapeutic path. The
aim of this article is to show plastic surgeons how
to effectively collect the best available evidence,
how to determine whether the collected evidence
is valid, and how to apply this evidence to indi-
vidual patients.
CLINICAL SCENARIO
You are a consulting plastic surgeon asked to
see a 32-year-old man who suffered an open frac-
ture dislocation of the right ankle joint after a
motor vehicle accident. He is otherwise healthy
with no comorbidities or other injuries from the
accident. On examination, he has a nearly cir-
cumferential soft-tissue defect around the ankle
with moderate wound contamination (Fig. 1). The
right dorsalis pedis is not palpable; however, the
foot is warm and appears to be well perfused.
The capillary refill of the toes is less than 2 sec-
onds. The patient has minimal flexion and exten-
sion of the toes of his right foot. He has no active
motion at the ankle; however, sensation in the
right foot is intact except for decreased sensation
in the superficial peroneal region and saphenous
nerve distribution. The patient is diagnosed with
a Gustilo type IIIB open tibial fracture.
You know that there are two treatment options
for this type of severe tibial fracture: (1) recon-
struction (limb salvage) involving reduction, fix-
ation, and microvascular free tissue transfer; and
(2) primary amputation. Generally, reconstruc-
tion is the preferred choice of surgeons, despite
the lack of evidence indicating better outcomes
associated with reconstruction.2In addition, both
the patient and his family desire to salvage the
injured limb. For these reasons, you recommend
salvage. The patient and his family are happy to
hear your suggestion and agree to the recon-
struction. However, as you discuss the proce-
From the Department of Plastic, Reconstructive, and Aes-
thetic Surgery, Nippon Medical School; the Department of
Hand and Reconstructive Microsurgery, National Univer-
sity Health System; and the Section of Plastic Surgery, De-
partment of Surgery, University of Michigan Health System.
Received for publication January 26, 2012; accepted Sep-
tember 14, 2012.
Copyright ©2013 by the American Society of Plastic Surgeons
DOI: 10.1097/PRS.0b013e31827c6f90
Disclosure: The authors have no financial interest
to declare in relation to the content of this article.
www.PRSJournal.com 655
dure and possible complications with the pa-
tient, he has a few questions: “It seems like
reconstruction has a lot of potential complica-
tions such as osteomyelitis, nonunion, flap fail-
ure, and subsequent secondary amputation.
Doctor, how long will it be before I can go back
to work? Of course, I would like to save my leg
if possible, but I cannot get off work for very
long. I just got a job as a construction worker and
I need to work. Can you tell me which surgery
would get me back to work faster?”
After the discussion, the patient realizes that
salvage is not always the best choice. Now, he wants
to know what to expect going forward after each
treatment option, and wants to learn more about
predicting outcomes, particularly recovery time
and the probability of secondary amputation
after reconstruction. You suddenly realize that
you are not sufficiently prepared to address
these questions.
WHAT IS A PROGNOSTIC STUDY?
To provide useful information for this patient
to make an informed choice regarding his treat-
ment, one must be aware of the existence of sev-
eral variables (e.g., patient characteristics, injury
severity, treatment timing) that can predict when
patients will do well and when they will do poorly.
These variables are known as prognostic factors.
Prognostic factors influence outcomes of condi-
tions that already exist or interventions to treat
those conditions (staging of lung cancer or injury
severity score following motor vehicle accidents)
(Table 1).3–8 Prognostic factors should be distin-
guished from risk factors that are associated with
creating a condition or new cases (such as smoking
causing lung cancer or drunk driving causing mo-
tor vehicle accidents). Therefore, risk factors are
characteristics causing the condition or disease,
whereas prognostic factors determine whether the
patient with the condition or disease will do well or
do poorly as determined by examining characteris-
tics of the clinical course following diagnosis or
injury. In addition, risk factors are determined by
examining the behaviors, characteristics, and expo-
sures that preceded a disease or injury. Prognostic
studies examine these factors to determine their re-
lationship to possible outcomes following various
therapeutic interventions and to predict the proba-
bility with which those outcomes can be expected to
occur.9
Like all studies, prognosis studies vary in the
level of evidence they provide. A hierarchical rat-
ing system of evidence offers important guidance
for the confidence one can place in the results of
a given study (Table 2). Randomized controlled
trials stand at the top of the hierarchy and are most
likely to provide unbiased evidence. However,
conducting a randomized trial of prognostic fac-
tors can be difficult, even impossible. Factors such
as patient characteristics and injury severity can-
not be randomized, and it may be unethical to
randomize patients to receive allocation of prog-
nostic factors such as extended ischemia time or
delayed de´bridement.10–12 For these reasons,
surgical studies investigating prognostic factors
are mainly observational studies, and the best
study design is a prospective cohort study (prog-
nostic level I evidence). An alternative observa-
tional study is a retrospective cohort study or a
retrospective case-control study; however, these
retrospective studies are of lower evidence level
compared with prospective studies. In the hier-
archy of evidence, case series are considered
level IV evidence. Case series do not have a
comparison group, and this makes data derived
from case series more prone to bias, specifically,
selection bias.
Fig. 1. A 32-year-old man with type IIIB open fracture dislocation of the
right ankle.
Plastic and Reconstructive Surgery March 2013
656
HOW TO PREDICT PROGNOSIS
Step 1: Formulate a Searchable Question
It can be difficult to keep up with the rapid
accumulation of medical and surgical literature. A
well-designed search, however, can cut through
the deluge of publications. The Patients, Inter-
vention, Comparison, and Outcomes methodol-
ogy enables one to identify the most relevant
sources and leads to a more efficient literature
search.13 To benefit patients and clinicians, clini-
cal questions need to be both directly relevant to
patients’ problems and phrased in ways that direct
the search to relevant and precise answers. In prac-
tice, well-built clinical questions contain four el-
ements, summarized below (Table 3):
Patients (P): What Is the Sample of
Interest?
Specifically, one must consider patient char-
acteristics that may affect outcome such as frac-
ture classification (e.g., type IIIB/IIIC open tib-
ial fractures), age (adult, 18 to 60 years old), sex,
medical history, general health, or concomit-
ant medications.
Intervention (I): What Are the Prognostic
Factors (or Interventions) Being Considered?
Studies answering specific clinical questions
can be classifieds into three main types: therapeu-
tic, prognostic, and diagnostic. Prognostic studies
focus on finding factors (e.g., patient character-
istics, treatment choice, presence or absence of
complications) that are associated with favorable/
unfavorable outcomes.
Comparison (C): What Is the Comparison
Treatment?
Your question sometimes will not require a
control group, such as when you are interested in
prognostic factor after one treatment option. If
you were examining interventions, in contrast, you
must define the control group (primary amputa-
tion) to which the intervention group (reconstruc-
tion with flaps) will be compared.
Outcomes (O): In What Outcomes Are You
and Your Patient Interested?
Many factors influence outcomes following
major limb trauma, and the relationships between
these factors are multivariate and complex.14
Thus, you must decide on what outcomes are im-
portant to the question. It is good to be specific
and to aim for the outcomes that are most impor-
tant to your patient (in our scenario, outcomes
should be limb function, complications, and time
taken to return to work).
Keeping all these factors in mind, the relevant
clinical question for our example can be formu-
Table 1. Examples of Risk Factors/Prognostic Factors in Plastic Surgery Field
Condition/Intervention Risk Factors Prognostic Factors
Malignant melanoma* Genetic predisposition Classification according to
histopathologic type
Skin type (Caucasian population
are at higher risk) Breslow and Clark classification
Presence of certain pigmented changes
on the skin (freckles, dysplastic nevus)
Mitotic index on mm2
Overexposure to UV light, sunburn in
early childhood and type of work
Presence or absence of ulceration
No. of metastatic lymph nodes and
presence of satellite metastases
Diabetic foot† Age (risks increase with age) Sex (men have poorer outcomes)
Sex (men are at higher risk) Delay of management
Race (African Americans, Hispanic Americans,
and Native Americans are at higher risk)
Quality of medical treatment
Duration of diabetes
Surgical attitude
Other complications from diabetes
Inadequate level of amputation
Poor blood glucose control
Smoking
Lack of structured prevention (patient
education, general practitioners’ training,
and cooperation between surgeons
and diabetologists)
Free-flap for head and
neck reconstruction‡
Diabetes Operation time (exceeding 10 hr)
Preoperative irradiation
*From Balc CM, Houhton AN, Peters LJ. Cutaneous melanoma. In: DeVita VT Jr, Hellman S, Rosenberg SA, eds. Cancer: Principles and Practice
of Oncology. 5th ed. Philadelphia: Lippincott; 1997:1612–1661; and Retsas S, Henry K, Mohammed MQ, MacRae K. Prognostic factors of
cutaneous melanoma and a new staging system proposed by the American Joint Committee on Cancer (AJCC): Validation in a cohort of 1284
patients. Eur J Cancer 2002;38:511–516.
†From Merza Z, Tesfaye S. The risk factors for diabetic foot ulceration. Foot 2003;13:125–129; and Benotmane A, Mohammedi F, Ayad F, Kadi
K, Azzouz A. Diabetic foot lesions: Etiologic and prognostic factors. Diabetes Metab. 2000;26:113–117.
‡From Joo YH, Sun DI, Park JO, Cho KJ, Kim MS. Risk factors of free flap compromise in 247 cases of microvascular head and neck
reconstruction: A single surgeon’s experience. Eur Arch Otorhinolaryngol. 2010;267:1629–1633; and Rosenberg AJ, Van Cann EM, van der Bilt
A, Koole R, van Es RJ. A prospective study on prognostic factors for free-flap reconstructions of head and neck defects. Int J Oral Maxillofac
Surg. 2009;38:666 670.
Volume 131, Number 3 Evidence on Prognosis
657
Table 3. Summary of the PICO Method
Method Prognostic Therapeutic Diagnostic
Patients/population (P) What patient group? What patient group? What patient group?
Type IIIB/IIIC open tibial
fractures
Same as prognostic Tibial fractures
Age, 18–60 yr
Intervention (I) (e.g.,
cause, prognostic
factor, treatment)
Are there any prognostic factors
associated with outcomes? What are the interventions? What diagnostic procedure?
Patient characteristics,
treatment choice
(reconstruction or
amputation), possible events
after each treatment choice
Reconstruction Gustilo classification
Comparison (C) (if
necessary) Are there any prognostic factors
not associated with outcome? What is the comparison
treatment? Is there any other procedure?
N/A (usually unnecessary) Amputation AO classification
Outcomes (O) In which outcomes are you
interested? In what outcomes are you
interested? In what outcomes are you
interested?
Function, complications,
satisfaction, recovery time, etc.
Same as prognostic Accurate outcome prediction
PICO, Patients, Intervention, Comparison, and Outcomes.
Table 2. Levels of Evidence by Category of Study*
Level
Prognostic Studies (Investigating
the Outcome of Disease)
Therapeutic Studies (Investigating
the Results of Treatment)
Diagnostic Studies (Investigating
a Diagnostic Test)
I 1. Systematic reviews† of level
I studies‡ 1. High-quality RCT 1. Testing of previously developed
diagnostic criteria in series of
consecutive patients (with
universally applied criterion
standard)
2. Prospective cohort study‡
a. Significant difference
2. Systematic review of level I studies‡
b. No significant difference, yet
narrow CI
2. Systematic review† of level I RCTs
(with homogenous results)
II 1. Retrospective cohort study1. Lesser quality RCT (follow-up less
than 80%, no blinding, improper
randomization)
1. Development of diagnostic
criteria on basis of consecutive
patients (with universally applied
criterion standard)
2. Lesser quality prospective cohort
study (e.g., patients enrolled at
different points in their disease
or less than 80% follow-up)
2. Prospective comparative cohort
study 2. Systematic review of level II
studies‡
3. Study of untreated controls from
a previous RCT 3. Systematic review of
4. Systematic review of level II
studies‡
a. Level II studies
b. Level I studies with
nonhomogenous results
III Case control study 1. Case-control study 1. Study of nonconsecutive patients
(no criterion standard)
2. Retrospective comparative
cohort study
2. Systematic review of level III
studies†
3. Systematic review of level III
studies
IV Case series (no or historical
controls) Case series (no or historical
controls) 1. Case control study
2. Poor reference standard
V Expert opinion Expert opinion Expert opinion
RCT, randomized controlled trial; CI, confidence interval.
*Adapted from the Center for Evidence-Based Medicine Web site (www.cebm.net).
†A study of results from two or more previous studies.
‡All patients enrolled at same point in the disease course (inception cohort) with greater than 80% follow-up.
§Patients compared with a control group of patients treated at the same time and institution.
Study initiated after treatment was performed.
¶Patients with a particular outcome compared to those without (failed vs. successful operations) and reviewed for exposure, and so forth.
Plastic and Reconstructive Surgery March 2013
658
lated as follows: Which treatment option (limb
salvage or primary amputation) will allow a young
adult with a type IIIB/ IIIC open tibial fracture to
return to work earlier?
Step 2: Perform a Literature Search
After your discussion with the patient and his
family, you consult the literature to find evidence
that provides prognostic information on open tib-
ial fractures. Available online sources can be
broadly categorized into those that focus on pri-
mary sources (e.g., randomized controlled trials,
observational studies, or case series) and those
that focus on secondary sources that provide syn-
thesis or synopsis of primary sources, such as prac-
tice guidelines or systematic reviews (Table 4).
For a quick and effective literature search, sec-
ondary sources can provide a good starting point.
If you can find appropriate guidelines or system-
atic reviews closely matching your clinical ques-
tion, you can use that evidence in your clinical
decision-making because the evidence has been
preappraised. By using the preappraised evidence,
a busy physician can avoid having to conduct an
independent systematic review of the literature
each time a clinical question is generated, and
consequently save valuable time. If searching sec-
ondary sources proves unfruitful, you must turn to
the primary sources. The Clinical Queries func-
tion in PubMed (Fig. 2) is an important addition
to the PubMed toolbox. This function represents
validated sets of study methodology filters that
allow the user to limit the search to the highest
level evidence for the five most important types of
study categories (therapy, diagnosis, prognosis,
etiology, and clinical prediction rules) (Fig. 3).
For each category, a narrow (specific) and a broad
(sensitive) search option is offered. In addition, a
separate search tool for finding systematic reviews
is provided.
In our example, using these databases, the com-
bined Patients, Intervention, Comparison, and Out-
comes terms with their synonyms “(tibial fracture*
OR tibia fracture* OR lower extremity trauma OR
lower leg injury* OR limb-threatening) AND (IIIB
OR IIIC) AND (limb salvage OR flap* OR amputa-
tion) AND (complication* OR prognostic* OR risk
OR Outcome* OR Predictor*)” generated 49 results
in the clinical study categories and four systematic
reviews in the systematic reviews section. A quick
reading of the titles and abstracts identifies 14 clin-
ical studies15–28 and two systematic reviews29,30 that
can potentially be useful for our clinical question.
Two systematic reviews29,30 contained seven clinical
studies16–18,20–22,27 that were duplicates within the 14
clinical studies; thus, we excluded these seven arti-
cles at this stage, leaving us with seven clinical
studies15,19,23–26,28 and two systematic reviews.29,30
Step 3: Ensure Internal Validity (How Well Is
Bias Limited in the Study Methodology?)
What Is Internal Validity?
The data from primary sources derived from
the literature search cannot be used directly for
clinical decision-making. First, you must appraise
them to determine their validity and applicability.
The key questions in the critical appraisal of an
article on prognosis are listed in Table 5. The first
step of the critical appraisal process is to deter-
mine to what extent the study is valid and, there-
fore, is likely to represent the “truth” (Table 5).
Internal validity is the confidence that can be
placed in the cause-and-effect relationship pre-
sented by a study. The internal validity of a study
can be assessed by using the following checklist:
Is the Sample Representative and
Homogeneous?
It would be ideal to study the entire population
with a particular disease. This is often not feasible,
and a representative sample of patients is usually
studied. It is important for this study sample to
clearly define and be representative of what is
normally seen in practice, so that the results from
the study can be generalized. The diagnostic cri-
teria, inclusion/ exclusion criteria, and demo-
graphic and disease-specific factors (disease sever-
ity, e.g., Gustilo grade IIIB/IIIC) should be clear.
Patients should also enter the study at the same
Table 4. Available Sources*
Primary sources
Citation databases
MEDLINE (www.ncbi.nlm.nih.gov/PubMed)
EMBASE (www.embase.com)
CINHAL (www.ebscohost.com/cinahl)
Bibliographies
Traditional review articles/textbooks/conference
proceedings/reviewed studies
Personal communication
Secondary sources
Guidelines
National Guidelines Clearinghouse (http://www.
guideline.gov/)
AAOS Evidence-Based Clinical Practice Guidelines (http://
www.aaos.org/Research/guidelines/guide.asp)
Structured abstracts
EBM Online (http://ebm.bmj.com)
ACP Journal Club (http://acpjc.acponline.org/shared/
menu_about.html)
Systematic reviews
Cochrane Library (http://summaries.cochrane.org)
*To search several of the databases simultaneously, you can use the
TRIP Database (www.tripdatabase.com/).
Volume 131, Number 3 Evidence on Prognosis
659
Fig. 3. PubMed Clinical Queries page. Type your search terms in the Search box and select your question Category (yellow) and
Scope (red). A search for Systematic Reviews runs at the same time as the Clinical Studies Categories search. You can preview
the first five citations of the results for each research area (blue).
Fig.2. PubMed Clinical Queries. Toaccess PubMed Clinical Queries,click on the ClinicalQueries link below
PubMed Tools on the main search page (note the pointing finger).
Plastic and Reconstructive Surgery March 2013
660
point in the course of the disease so that the prog-
nostic risk is similar.31 For example, in our clinical
scenario, only studies that treated open tibial frac-
tures as a result of acute trauma were included.
Five studies were excluded because they included
patients with old injuries (2 months after injury)
or osteomyelitis.23–25,28,29
Are Measured Outcomes Unbiased and
Objective?
Authors should clearly define their target out-
comes based on the most clinically relevant mea-
sure. This is sometimes difficult. For example, one
article identified in our search defined union as
“full, unprotected weight-bearing after which no
further skeletal complication or reoperation
occurred,”32 whereas another defined nonunion
as “motion at the site of the fracture on manual
manipulation and no evidence of healing as seen
on roentgenograms that were made six months
after the original injury.”33 Because these subjec-
tive outcomes provide important information on
prognosis, physicians should strive to include
these factors to assess the outcomes using previ-
ously validated and reliable scales such as the
Lower Extremity Functional Scale.34
Was Patient Follow-Up Long Enough and
Complete?
Follow-up should be long enough to develop
a valid picture of the extent of the outcome of
interest. The period that constitutes “long
enough” will vary depending on the outcome. Loss
to follow-up is important, and only studies that
report this number should be used for decision-
making. The “5 and 20” rule is useful to determine
whether loss to follow-up is reasonable. If less than
5 percent of the study sample is lost to follow-up,
one can be assured that the loss has minimally
impacted the results. If, however, more than 20
percent of the study sample is lost to follow-up,
one should be cautious when making clinical de-
cisions based on study findings.35,36
Of the seven clinical articles and two system-
atic reviews that were examined for internal va-
lidity, four clinical articles and one systematic re-
view were excluded because the sample was not
representative. We are now left with three clinical
articles15,19,26 and one systematic review30 at the
end of step 3.
Step 4: What Are the Results?
How Likely Are the Outcomes over Time?
If the study appears to be valid, you may then
extract information regarding the effect of prog-
nostic factors on outcomes. Surgical studies mon-
itoring outcomes should report on the number of
favorable or unfavorable events over a certain pe-
riod, which can be used to create Kaplan-Meier
curves.37 Every time a patient has an outcome, the
ratio is recalculated. Using these calculations, a
curve can be generated that graphically depicts
the probability of return to work as time passes.
Figure 4 shows the percentage of patients who
return to work based on the chosen treatment. In
patients who underwent limb salvage, 63.5 percent
returned to work, and the mean delay before re-
turning to work was 14.4 months. In contrast, in
patients who underwent primary amputation, 73
percent returned to work, and the mean delay
before returning to work was 13.6 months.30 Out-
comes are often reported simply as a rate (e.g.,
rate of secondary amputation after limb salvage is
7.3 percent); however, rates convey very little in-
formation, and there can be important differences
in prognosis within similar summary rates. For
these reasons, Kaplan-Meier curves are used to
estimate outcomes of a cohort over time. It is a
useful method for describing any dichotomous
outcome that occurs only once during the fol-
low-up period.
How Precise Are the Prognostic Estimates?
A prognostic study provides only an estimate
of the true risk. Therefore, you need to judge how
precise the estimates in the study are. The CI gives
a measure of the precision (or uncertainty) of
study results when applied to the population of all
such patients. The 95 percent CI represents the
range of values within which we can be sure that
the true value lies 95 percent of the time. The
narrower the CI, the more precise the estimate. CI
is affected by sample size and by variability among
subjects. The larger the trial’s sample size or the
larger the number of outcome events is, the nar-
rower the confidence interval, and the “precision”
is increased. Because we excluded no articles
Table 5. Evaluating an Article on Prognosis
Ensure the study methods (internal validity)
Is the sample representative and homogeneous?
Are measured outcomes unbiased and objective?
Is patient follow-up long enough and complete?
What are the results?
How likely are the outcomes over time?
How precise are the prognostic estimates?
Ensure applicability of the study to your own patient
(external validity)
Do the study patients match your own patients and was
management similar?
Was the follow-up long enough?
Can the study results be used?
Volume 131, Number 3 Evidence on Prognosis
661
based on step 4, we are now left with three clinical
articles17,19,26 and one systematic review.30
Step 5: Ensure External Validity
What Is External Validity?
Once you ensure internal validity of the study
and understand its results, a remaining question is
whether you can apply the results to your own
patients, known as external validity (Fig. 5). To
evaluate how well you can generalize the results
from a study to your own patient, you must ask
yourself the following two questions:
Is My Patient Similar to Those in the Study?
Although it is important to specifically define
your patient sample before performing a litera-
ture search, it may not be possible to find a study
with that exact sample. Comparison between your
patients and those in the article, using descrip-
tions of the study sample’s demographic and clin-
ical characteristics, is necessary. Careful interpre-
tation of subgroup analyses is important as well. In
our example, one study was excluded because the
study included not only tibial fractures but also
foot fractures, and did not describe subgroup
information in detail.19 Furthermore, treatment
options used in the study should be similar, if
not identical, to those that you can provide. We
excluded one article26 in which the study sample
was treated primarily with vascularized bone
grafts because they did not represent the usual
reconstructive method of primary skin or mus-
Fig. 4. An example of Kaplan-Meier curves for two conditions (limb salvage and am-
putation) associated with rates of patients who return to work.
Fig. 5. Internal and external validity. The blue ellipse represents internal
validity, and the red rounded ellipse around it represents external valid-
ity. Internal validity is generally a prerequisite for external validity.
Plastic and Reconstructive Surgery March 2013
662
cle flap coverage followed by secondary bone
grafting.
How Will the Results Impact My Patient
Care?
Prognostic data often reveal certain factors
that can significantly influence the outcome of a
proposed treatment. Knowing the expected clin-
ical course of your patient can help you to judge
what treatment should be offered. Even if the
prognostic data do not lead to a clearly superior
treatment option, they can still be clinically useful.
For example, the current literature offers no ev-
idence to support superior final outcomes of ei-
ther limb salvage or primary amputation for
Gustilo type IIIB/IIIC tibial fractures.30,38 How-
ever, you can provide useful outcome data for
each treatment course, such as recovery time and
possible complications during the treatment
course.
In addition, beyond the prescription of an ef-
fective therapy, prognostic data can be clinically
useful when reassuring your concerned patient
and his family as they face an uncertain future.
Unfortunately, salvaging a severely injured limb
does not substantially improve the patient’s out-
comes of interest (primarily, return to work time)
compared with primary amputation. However,
even this inconclusive prognostic information can
assist the patient and his family in making impor-
tant quality-of-life decisions and planning the next
part of their lives after the injury. At this stage, we
excluded one article17 because the article did not
measure the objective outcomes that our patient
wanted to know, such as recovery time. As a result,
three clinical articles17,19,26 were excluded at the
end of step 5.
Step 6: Resolve Your Clinical Scenario
Finally, one systematic review by Saddawi-
Konefka et al.30 remained to answer our patient’s
question (Table 6). Using the study, you can re-
turn to your patient and his family and confidently
inform them that, following limb salvage, it takes
approximately 8 months until full weight-bearing
and approximately 60 percent of patients return to
work at approximately 14 months after injury. These
figures are each slightly better in patients who have
had primary amputation, but are not significantly
different. Secondary amputation, as a complication
after leg salvage, occurs in approximately 6 to 10
percent, and the prognostic factors of secondary
amputation are infection and flap loss.
After considering this, the patient ultimately de-
cided to undergo limb salvage. The patient and his
Table 6. Characteristics of Eligible Study
Reference Study Design Sample Follow-Up Measured Outcomes Results
Saddawi-Konefka
et al., 200830 Systematic review Total 28 studies N/A Length of hospital stay Length of hospital stay, 56.9 days for
salvage patients and 63.7 days
for amputees
Patients with IIIB/
IIIC OTfx Complications Complications after limb salvage include
osteomyelitis (17.9%), nonunion (15.5%),
secondary amputation (7.3%), flap
failure (5.8%)
All acute trauma cases Rehabilitation time Rehabilitation time for salvaged patients
was reported as time to union (10.2 mo)
and time to full weight-bearing (8.1 mo)
Mean age, 22–48 yr QOL Pain, QOL, and limb function outcomes
were assessed differently among studies
and could not be combined
Limb function
Pain
Return-to-work data
The proportion of patients who returned
to work was 63.5% for salvage patients and
73% for amputees
N/A, not available; OTfx, open tibia fractures; QOL, quality of life.
Volume 131, Number 3 Evidence on Prognosis
663
family were disappointed to hear the poor outcomes
after limb salvage operations and longer recovery
time than they expected. However, the prognostic
information was reassuring; the predicted outcomes
allowed him to be proactive in planning the next
phase of his life after this devastating injury.
Kevin C. Chung, M.D., M.S.
Section of Plastic Surgery
University of Michigan Health System
2130 Taubman Center, SPC 5340
1500 East Medical Center Drive
Ann Arbor, Mich. 48109-5340
kecchung@umich.edu
ACKNOWLEDGMENTS
This work was supported in part by grants from the
National Institute on Aging and the National Institute
of Arthritis and Musculoskeletal and Skin Diseases (R01
AR062066 and 2R01 AR047328-06) and by a Mid-
career Investigator Award in Patient-Oriented Research
(K24 AR053120) from the National Institute of Arthri-
tis and Musculoskeletal and Skin Diseases (to K.C.C.).
REFERENCES
1. Merrill RM. Clinical epidemiology. In: Introduction to Epide-
miology. 5th ed. Sudbury, Mass: Jones and Bartlett; 2010:299–
311.
2. Parrett BM, Matros E, Pribaz JJ, Orgill DP. Lower extremity
trauma: Trends in the management of soft-tissue reconstruc-
tion of open tibia-fibula fractures. Plast Reconstr Surg. 2006;
117:1315–1322; discussion 1323–1324.
3. Balc CM, Houhton AN, Peters LJ. Cutaneous melanoma. In:
DeVita VT Jr, Hellman S, Rosenberg SA, eds. Cancer: Principles
and Practice of Oncology. 5th ed. Philadelphia: Lippincott;
1997:1612–1661.
4. Retsas S, Henry K, Mohammed MQ, MacRae K. Prognostic
factors of cutaneous melanoma and a new staging system
proposed by the American Joint Committee on Cancer
(AJCC): Validation in a cohort of 1284 patients. Eur J Cancer
2002;38:511–516.
5. Merza Z, Tesfaye S. The risk factors for diabetic foot ulcer-
ation. Foot 2003;13:125–129.
6. Benotmane A, Mohammedi F, Ayad F, Kadi K, Azzouz A.
Diabetic foot lesions: Etiologic and prognostic factors.
Diabetes Metab. 2000;26:113–117.
7. Joo YH, Sun DI, Park JO, Cho KJ, Kim MS. Risk factors of
free flap compromise in 247 cases of microvascular head
and neck reconstruction: A single surgeon’s experience.
Eur Arch Otorhinolaryngol. 2010;267:1629–1633.
8. Rosenberg AJ, Van Cann EM, van der Bilt A, Koole R, van Es
RJ. A prospective study on prognostic factors for free-flap
reconstructions of head and neck defects. Int J Oral Maxillofac
Surg. 2009;38:666–670.
9. Bhandari M, Guyatt GH, Swiontkowski MF. User’s guide to
the orthopaedic literature: How to use an article about prog-
nosis. J Bone Joint Surg Am. 2001;83:1555–1564.
10. McCulloch P, Taylor I, Sasako M, Lovett B, Griffin D. Ran-
domised trials in surgery: Problems and possible solutions.
BMJ. 2002;324:1448–1451.
11. McLeod RS. Issues in surgical randomized controlled trials.
World J Surg. 1999;23:1210–1214.
12. Demange MK, Fregni F. Limits to clinical trials in surgical
areas. Clinics (Sao Paulo) 2011;66:159–161.
13. Richardson WS, Wilson MC, Nishikawa J, Hayward RS. The
well-built clinical question: A key to evidence-based deci-
sions. ACP J Club 1995;123:A12–A13.
14. MacKenzie EJ, Bosse MJ. Factors influencing outcome fol-
lowing limb-threatening lower limb trauma: Lessons learned
from the Lower Extremity Assessment Project (LEAP). JAm
Acad Orthop Surg. 2006;14:S205–S210.
15. Giannoudis PV, Harwood PJ, Kontakis G, et al. Long-term
quality of life in trauma patients following the full spectrum
of tibial injury (fasciotomy, closed fracture, grade IIIB/IIIC
open fracture and amputation). Injury 2009;40:213–219.
16. Tielinen L, Lindahl JE, Tukiainen EJ. Acute unreamed in-
tramedullary nailing and soft tissue reconstruction with mus-
cle flaps for the treatment of severe open tibial shaft frac-
tures. Injury 2007;38:906–912.
17. Rajasekaran S, Naresh Babu J, Dheenadhayalan J, et al. A
score for predicting salvage and outcome in Gustilo type-IIIA
and type-IIIB open tibial fractures. J Bone Joint Surg Br. 2006;
88:1351–1360.
18. Pelissier P, Boireau P, Martin D, Baudet J. Bone reconstruc-
tion of the lower extremity: Complications and outcomes.
Plast Reconstr Surg. 2003;111:2223–2229.
19. MacKenzie EJ, Bosse MJ, Kellam JF, et al. Factors influencing
the decision to amputate or reconstruct after high-energy
lower extremity trauma. J Trauma 2002;52:641–649.
20. Dagum AB, Best AK, Schemitsch EH, Mahoney JL, Mahomed
MN, Blight KR. Salvage after severe lower-extremity trauma:
Are the outcomes worth the means? Plast Reconstr Surg. 1999;
103:1212–1220.
21. O’Sullivan ST, O’Sullivan M, Pasha N, O’Shaughnessy M,
O’Connor TP. Is it possible to predict limb viability in com-
plex Gustilo IIIB and IIIC tibial fractures? A comparison of
two predictive indices. Injury 1997;28:639–642.
22. Gopal S, Majumder S, Batchelor AG, Knight SL, De Boer P,
Smith RM. Fix and flap: The radical orthopaedic and plastic
treatment of severe open fractures of the tibia. J Bone Joint
Surg Br. 2000;82:959–966.
23. Khouri RK, Shaw WW. Reconstruction of the lower extremity
with microvascular free flaps: A 10-year experience with 304
consecutive cases. J Trauma 1989;29:1086–1094.
24. Laughlin RT, Smith KL, Russell RC, Hayes JM. Late func-
tional outcome in patients with tibia fractures covered with
free muscle flaps. J Orthop Trauma 1993;7:123–129.
25. Lin CH, Wei FC, Levin LS, Su JI, Yeh WL. The functional
outcome of lower-extremity fractures with vascular injury.
J Trauma 1997;43:480–485.
26. Musharrafieh R, Osmani O, Saghieh S, Elhassan B, Atiyeh B.
Microvascular composite tissue transfer for the management
of type IIIB and IIIC fractures of the distal leg and compound
foot fractures. J Reconstr Microsurg. 1999;15:501–507.
27. Redett RJ, Robertson BC, Chang B, Girotto J, Vaughan T.
Limb salvage of lower-extremity wounds using free gracilis
muscle reconstruction. Plast Reconstr Surg. 2000;106:1507–
1513.
28. Yajima H, Tamai S, Kobata Y, Murata K, Fukui A, Takakura
Y. Vascularized composite tissue transfers or open fractures
with massive soft-tissue defects in the lower extremities. Mi-
crosurgery 2002;22:114–121.
29. Papakostidis C, Kanakaris NK, Pretel J, Faour O, Morell DJ,
Giannoudis PV. Prevalence of complications of open tibial
shaft fractures stratified as per the Gustilo-Anderson classi-
fication. Injury 2011;42:1408–1415.
Plastic and Reconstructive Surgery March 2013
664
30. Saddawi-Konefka D, Kim HM, Chung KC. A systematic review
of outcomes and complications of reconstruction and am-
putation for type IIIB and IIIC fractures of the tibia. Plast
Reconstr Surg. 2008;122:1796–1805.
31. Hansebout RR, Cornacchi SD, Haines T, Goldsmith CH. How to
use an article about prognosis. Can J Surg. 2009;52:328–336.
32. Hertel R, Lambert SM, Mu¨ ller S, Ballmer FT, Ganz R. On the
timing of soft-tissue reconstruction for open fractures of the
lower leg. Arch Orthop Trauma Surg. 1999;119:7–12.
33. Caudle RJ, Stern PJ. Severe open fractures of the tibia. J Bone
Joint Surg Am. 1987;69:801–807.
34. Binkley JM, Stratford PW, Lott SA, Riddle DL. The Lower
Extremity Functional Scale (LEFS): Scale development, mea-
surement properties, and clinical application. North Amer-
ican Orthopaedic Rehabilitation Research Network. Phys
Ther. 1999;79:371–383.
35. Fewtrell MS, Kennedy K, Singhal A, et al. How much loss to
follow-up is acceptable in long-term randomised trials and
prospective studies? Arch Dis Child. 2008;93:458–461.
36. Schulz KF, Grimes DA. Sample size slippages in randomised
trials: Exclusions and the lost and wayward. Lancet 2002;359:
781–785.
37. Rich JT, Neely JG, Paniello RC, Voelker CC, Nussenbaum B,
Wang EW. A practical guide to understanding Kaplan-Meier
curves. Otolaryngol Head Neck Surg. 2010;143:331–336.
38. Bosse MJ, MacKenzie EJ, Kellam JF, et al. An analysis of
outcomes of reconstruction or amputation after leg-threat-
ening injuries. N Engl J Med. 2002;347:1924–1931.
Recent CME Articles
See the May 2011 CME article by Fox and
Mackinnon. Go online today at PRSJournal.com
to see the article and take the CME test for 1
hour of Category 1 credit.
Volume 131, Number 3 Evidence on Prognosis
665
  • [Show abstract] [Hide abstract] ABSTRACT: A study was performed to analyze the results and final outcomes of bone reconstruction of the lower extremity. Twenty-six patients presented with type IIIB open fractures, nine with type IIIC open fractures, and 15 with chronic osteomyelitis. Seven patients underwent primary amputation, and reconstruction was attempted for 43 patients. The mean bone defect size was 7.7 cm (range, 3 to 20 cm). Bone reconstruction was achieved with conventional bone grafts in 16 cases, in association with either local (13 cases) or free (three cases) flaps. Vascularized bone transfer was performed in 24 cases, with either osteocutaneous groin flaps (10 cases), soleus-fibula flaps (12 cases), or osteocutaneous lateral arm flaps (two cases). For three patients, bone reconstruction was performed with a technique that combines the induction of a membrane around a cement spacer with the use of an autologous cancellous bone graft. Infections were observed to be responsible for prolonged hospital stays and treatment failures. The cumulative rates of sepsis were 4.6 percent at 1 week after injury and 62.8 percent at 2 months. Vascular complications were also related to infections and were responsible for four secondary amputations. One patient asked for secondary amputation because of a painful nonfunctional lower limb. Bone healing occurred in 37 of 43 cases, and the average time to union was 9.5 months, with an average of 8.7 procedures. The mean lengths of stay were 49 days for conventional bone grafts and 62 days for vascularized bone grafts. All of the 50 patients were able to walk, with an average time of 14 months. All of the patients with amputations underwent prosthetic rehabilitation. Patients mostly complained about the reconstructed limb (62.8 percent). Joint stiffness was present in 40 percent of the cases. Other long-term complications were pain (nine cases), lack of sensation (five cases), infection (five cases), and pseudarthrosis (one case). However, all of the patients with successful reconstructions preferred their salvaged leg to an amputation. Of 41 patients who were working before the injury, 26 returned to work.
    Article · Jul 2003
  • [Show abstract] [Hide abstract] ABSTRACT: Advances in reconstructive surgery have allowed for impressive salvage after severe lower-extremity trauma but not without complications when compared with immediate below-knee amputation. Several amputation index scores have been developed to help predict successful salvage as defined by a viable rather than a functional extremity. The purpose of this study was to evaluate retrospectively the predictive value of the amputation index scores and to assess prospectively overall health status and specific dysfunction in successful limb salvage and primary and secondary amputation by administering standardized generic and specific outcomes questionnaires (Medical Outcomes Study 36-Item Short-Form Health Survey, Western Ontario and MacMaster Universities Osteoarthritis Index). A retrospective chart review identified 55 severe lower-extremity injuries (Gustilo Type IIIB and IIIC) over a 12-year period (1984 to 1996). Forty-six severe open tibial fractures in 45 patients underwent attempted salvage. All required soft-tissue coverage by either local or free flap or vascular repair for leg salvage. The attempted-salvage group was subdivided into successful salvage and secondary amputation. The other nine patients underwent a primary amputation. There were no statistically significant differences in terms of patient demographics or other injuries (Injury Severity Score) in the three groups. Forty-eight of 54 patients with an average 5-year follow-up completed a validated generic and specific outcomes health questionnaire. In the attempted-salvage group, 89 percent of patients had a successful salvage and 11 percent came to a secondary amputation. The amputation index scores correctly predicted an amputation in 32 percent of patients. The magnitude of the amputation index scores did not correlate with the physical outcomes scores and were not found to add any significant value of information to the surgeon's decision making. Patients undergoing primary and secondary amputation had a worse physical outcomes score (28 versus 38) than successful salvage (p < 0.007). Even so, the SF-36 (physical component score) outcomes score for this group of injured extremities, regardless as to whether salvaged or amputated, was as low as or lower than that of many serious medical illnesses, suggesting that severe lower-extremity trauma impairs health as much as or more than being seriously ill. The mental component score in this group was comparable to that of a healthy population (49 versus 50), which implies the disability is primarily physical rather than psychological. Ninety-two percent of patients preferred their salvaged leg to an amputation at any stage of their injury, and none would have preferred a primary amputation.
    Article · May 1999
  • [Show abstract] [Hide abstract] ABSTRACT: The Lower Extremity Assessment Project (LEAP) is a multicenter study of severe lower extremity trauma in the US civilian population. At 2- and 7-year follow-ups, the LEAP study found no difference in functional outcome between patients who underwent either limb salvage surgery or amputation. However, outcomes on average were poor for both groups. This study and others provide evidence of wide-ranging variations in outcome following major limb trauma, with a substantial proportion of patients experiencing long-term disability. In addition, outcomes often are more affected by the patient's economic, social, and personal resources than by the initial treatment of the injury--specifically, amputation or reconstruction and level of amputation. A conceptual framework for examining outcomes after injury may be used to identify opportunities for interventions that would improve outcomes. Because of essential differences between the civilian and military populations, the findings of the LEAP study may correlate only roughly with combat casualty outcomes.
    Article · Nov 2006
Show more