Fontan Cross-Sectional Study
Public Use Dataset
About the Study
The NHLBI Fontan Cross-Sectional Study was conducted by the Pediatric Heart Network
(PHN) at 7 centers in 2003-2004. The PHN screened a total 1,078 patients and
enrolled 546 children aged 6 to 18 years old. Study measurements were specified
to be made in a 3-month window following enrollment. The primary aim of the study
was to examine associations between functional health status (measured by parent-
and child-report questionnaires) and ventricular state and performance (measured
by 2D and Doppler echocardiography, maximal exercise testing, ECG, cardiac MRI
and resting B-type natriuretic peptide [BNP] concentration). Core laboratories were
used for interpretation of echocardiograms and MRIs, and to perform the BNP assay.
Test completion rates in the enrolled cohort ranged from 536 echocardiograms and
511 parent-report questionnaires to 161 MRIs that were acceptable for analysis.
The study design has been summarized in Sleeper at al. (AHJ 2006) and in the study
protocol (available to users with approved logins). Tables 1 and 2 provide selected
subgroup sizes for the Full Protocol Fontan Study cohort. A great deal of additional
information on available sample sizes for measurements may be found in Anderson et al.
(JACC 2008; http://www.pediatricheartnetwork.org/publications/Fontan_Mainresultspaper.pdf),
as well the published articles on specialized topics (see posted Bibliography at http://www.pediatricheartnetwork.org/pubFontan.asp).
Enrollment Age Distribution of the Fontan Study Cohort by Age at Fontan Surgery
|Age at Fontan, yr||Age at enrollment, yr|
Distribution of Cardiac Anatomic Diagnosis and Ventricular Morphology
|Pre-Fontan Cardiac Anatomic Diagnosis|
| ||A1.01: SV, DILV||A1.02: SV, DIRV||A1.03: SV, MA
||A1.04: SV, TA||A1.05: SV, Unbalanced AVCD||A1.06: SV, Heterotaxia syndrome
||A2: HLHS||A3: Other functional SV not fitting any other categories||Total|
SV=single ventricle; LV=left ventricular; DI=double inlet; MA=mitral atresia;
TA=tricuspid atresia; HLHS=hypoplastic left heart syndrome; AVCD=Atrioventricular canal defect
The following datasets and descriptor files are available for download. A login and password
(request access via http://www.pediatricheartnetwork.org)
are required for download capability. The lock date used for creation of the public dataset is
August 1, 2011. Privacy protection of these data is described in Appendix A.
- Annotated study data collection forms (PDF) — These contain the SAS variable names
next to each data field on the form. These form documents also include some created
variables and their definitions.
- SAS version 9.2 datasets
- The file fontanformats.sas7bcat — Include this file in your program using:
options fmtsearch = (fmtlib.fontanformats);
where fmtlib is specified using a libname statement as the path name.
- SAS Proc Contents for each dataset (PDF)
- Excel datasets (with variable formats applied) — These data have a .csv extension,
which means that the file may also be opened either in Excel, OR in a text editor,
appearing as a comma-delimited file.
Resources posted on the pediatricheartnetwork.org website include:
Data Use Policy
- REQUIRED ACKNOWLEDGEMENTS: All presentations and publications
using these data must include the following statement:
"The NIH/NHLBI Pediatric Heart Network Fontan Cross-Sectional
Study dataset was used in preparation of this work. Data were downloaded from
- PAPER, ABSTRACT, and PRESENTATION TITLES: Titles may, at the authors'
discretion, mention the PHN database but should not imply that the work is
from the PHN. An example of an acceptable phrase would be, "an analysis
of the Pediatric Heart Network public database." Whether or not the title
makes mention of the PHN, acknowledgement should be made as described above.
- All users are requested to send a copy of published abstracts and articles
to the PHN Data Coordinating Center at New England Research Institutes
within one month of publication. This will allow the PHN and the NHLBI
to document the continued impact of this study on the field.
- The login and password provided to each user are valid for 6 months.
If a user decides to complete analyses leading to more than one presentation
or publication in that time period, it is requested that they notify
the PHN Data Coordinating Center at New England Research Institutes
of their additional analysis topics, solely for the purposes of tracking.
- The login and password to access the public dataset is provided to a
single user. If a colleague would like to access the public dataset
for a different analysis topic, a separate request for login and password
should be submitted via the
Tips On Using These Data
- Identification numbers for study subjects and study sites
have been re-assigned for privacy protection.
subj_id: Subject ID ranging from 1 to 1078;
site_id: Ranges from 1 to 7
- The study data are contained in a large number of individual forms.
These may be used jointly by merging on subj_id. No dataset
has more than one record per subject. A single dataset of 168
of the most commonly used raw and created variables,
FXS_KEYINFOPUB, is also provided. It contains 1078 records.
- It is important to keep in mind that some forms include records
for more than the 546 Full Protocol subjects. Forms F001, F02A,
and F02B contain some screening records and the functional health
status forms contain records for the 60 Partial Protocol patients.
The Full Protocol cohort may be selected with consented=1
(found in FXS_KEYINFOPUB and F02B) or disp=7 (FXS_KEYINFOPUB).
- The raw data collected are contained in the original variables
(denoted by upper case variable names). Prior to analysis, these
variables must have special values (typically negative numbers, see
Appendix B) set to missing. Created variables (denoted by lower case
variable names) already contain a SAS missing value if the measurement
- To select for echocardiograms and MRIs that have data acceptable
for analysis, use ACPTECHO=1 and ACCPTMRI=1. Unacceptable
echocardiograms and MRIs have no qualitative or quantitative
- Anatomy: A key grouping variable that has been used for many
study analyses is ventricular morphology (left, right, mixed).
This variable is determined according to cardiac anatomic diagnosis
and is called vent_type (Form 001 and dataset FXS_KEYINFOPUB).
This variable is not the same as the echo core laboratory (Form 12B)
variable vent_dom, which does not take into account anatomic diagnoses
involving reversed location of cardiac structures. All Fontan Study
publications have utilized vent_type.
- The core laboratory echocardiographic measurement dataset Form F12B
contains many created variables that are most commonly used in analysis.
These variables express total ventricular size (e.g., echoedv)
and function (e.g., echoef) and overall regurgitation grades
(oavvregurg, slvregurg). They incorporate, where available,
measures from both ventricles, and measurements from left, right, and
common atrioventricular valves. Echocardiographic z-scores (e.g.,
echoedv_z, echoef_z) accounting for body surface area
or age that provide a reference relative to normal (two-ventricle) children
are also included in this dataset.
- Exercise testing was conducted in 411 subjects. This cohort is often divided
into those who did and did not achieve maximal effort. All Fontan Study
publications defined maximal effort as a respiratory exchange ratio
(peak respiratory quotient) ≥ 1.1. The variable on Form F10B called
MAXEFF (Question F1) was not used. The created variable rer_ge1_1
- Anthropometric z-scores are calculated using the 2000 CDC standard and
are stored in FXS_KEYINFOPUB. The raw measurements used for z-score
calculation are weight and height on Form F12B (Core Laboratory
- There has been one noticeable change in the study dataset that was made
after the majority of Fontan Study papers were published. The change is
with regard to the Type of Fontan Procedure. The corrected data have a
much lower proportion of extracardiac lateral tunnel procedures
(fontan_sgcat="B2.02.04") and a higher proportion of extracardiac
conduits (fontan_sgcat="B2.02.05"). The variable fontan_sgcat
is found in both the fxs_keyinfopub and f04bpub datasets. In the enrolled
cohort of 546 subjects, the frequencies are: 72 atriopulmonary connections;
327 total cavopulmonary connection (TCPC) intracardiac lateral tunnels;
3 TCPC extracardiac lateral tunnels; 133 TCPC extracardiac conduits;
and 11 other.
If you have questions about the study dataset that this documentation and
the above resources (protocol, articles) have not answered, please contact Lynn Sleeper
(firstname.lastname@example.org) or Victor Zak
at the PHN Data Coordinating Center (617-923-7747).
Implementation of Privacy Protection Rules for Public Use of the PHN Fontan Study Dataset
Variables that could lead to subject identification were eliminated in the public dataset. Steps included:
- Removal of original study ID number (replaced with subj_id,
a random consecutive numbering ranging from 1 to 1078),
and removal of acrostic. Of note, no names, addresses, zip code,
or medical record numbers were ever contained in the original study dataset.
- Seven centers contributed data to the Fontan Cross-Sectional Study.
A new center identifier (site_id), which represents a random
consecutive numbering ranging from 1 to 7 was created, without formats
(i.e., without center names).
- All dates in the original datasets were removed, and replaced with
"Age at event/intervention/procedure" in years (to 2 decimal places).
Therefore, time intervals may be calculated by subtraction of two ages.
The one exception to this convention is that the calendar year
of the most recent Fontan procedure performed remains in the dataset
for the 546 enrolled (Full protocol) subjects, in order for analyses
to adjust as needed for era effects. This variable is called
FONTAN_YEAR and is located in the Form 04B dataset and the
FXS_KEYINFOPUB dataset. Provision of a year only without day or
month conforms to HIPAA requirements.
- Free (write-in) text variables remain in the public dataset.
These often provide highly relevant information for interpretation
of the data. However, any write-in string that referred to a
specific date, a particular medical center or a particular MD
was blinded or omitted.
- Outliers for continuous variables and small group sizes for
categorical variables were retained in the dataset for public use
due to their importance in interpretation of the data and low
likelihood of unblinding any user to a subject identity unless
the user already had access to the particular medical
center's data for valid reasons.
Special Value Codes
-9 = missing
-8 = don't know/indeterminate
-7 = refused to answer
-6 = not recorded
-5 = measurement could not be reliably recorded or is not interpretable (study technically inadequate)
-4 = illegible
-2 = programmed skipped field based on results of or response to a previous question
-1 = not applicable/structure not present
-77 = Not detectable below 4.0 pg/ml (BNP concentration)