Biventricular repair or single-ventricle palliation: what the scores built for this lesion say, in one bottom line
Experimental prototype. Built from published sources but not yet independently verified. Do not use for clinical decisions; check results against the primary references.
All fields start blank. Choose the lesion; only the scores built for it count toward the bottom line.
CHSS models: single ventricle or biventricular repairNot calculated here
The Congenital Heart Surgeons' Society (CHSS) published two models that estimate, for one neonate with critical left ventricular outflow tract obstruction (LVOTO), how much 5-year survival differs between a single-ventricle pathway and biventricular repair. This page explains both but calculates neither. Their coefficients could not be checked against a readable copy of either paper; the details are under “Model, equations and sources”. Use the official CHSS tool or the papers.
CHSS 1: Lofland 2001, survival benefit
Cohort. 320 neonates with critical LV outflow obstruction, entered prospectively at several institutions from 1994 to 2000. Intended biventricular repair: 116 (70% 5-year survival). Norwood: 179 (60%).
Output. Predicted survival benefit of the Norwood procedure over biventricular repair. The authors fitted a linear regression to each patient's predictions from two separate hazard models, one per pathway. A positive value favours Norwood and a negative value favours biventricular repair (reviews; see sources).
Greater benefit from Norwood with younger age at entry, lower aortic valve and LV length z scores, a higher EFE grade, no important tricuspid regurgitation, and a larger ascending aorta (abstract). Reviews specify the aortic valve z at the sinuses and moderate or severe TR.
Cohort. 362 neonates from 26 institutions, 1994 to 2001: 223 had intended univentricular repair (84 deaths) and 139 biventricular repair (39 deaths). Critical LVOTO was stenosis anywhere from below the aortic valve to the innominate artery, with concordant AV and VA connections.
Output. Predicted 5-year survival after univentricular repair minus that after biventricular repair. “A positive value favors UVR management and a negative value favors BVR management.” R² 0.92.
Death after univentricular repair was more likely with moderate or severe TR, a smaller mitral annulus z, a shorter apex-forming ventricle (crux to apex, indexed to height) and a large VSD.
Death after biventricular repair was more likely with:
a smaller minimum LVOT diameter, measured anywhere from below the valve to the brachiocephalic artery, indexed to BSA and entered after inverse transformation;
a higher EFE grade (0 none; 1 papillary muscles only; 2 papillary muscle with some endocardial surface; 3 extensive endocardial surface);
any LV dysfunction, including mild;
a smaller mid-aortic arch, measured just proximal to the left subclavian artery and indexed to BSA.
Why it matters. When management went against the prediction, biventricular repair carried 30 observed deaths against 14 expected (P < .001), and univentricular repair 20 against 13 (P = .02).
Before you use either model
Inputs on this page cannot be reused. CHSS 1 uses the aortic valve z at the sinuses, not the annulus z that Colan uses, and both CHSS models grade EFE from 0 to 3, not yes/no for grade 2 or 3.
Neither paper names its z-score reference in what was readable here. Hickey 2007 says only “published normative data if available”.
Both cohorts are from 1994 to 2001. External checks in later single-centre cohorts found that many neonates predicted to need a single-ventricle route did well with biventricular repair (see sources).
Official tool: Hickey 2007 says the UVR-SA is freely available on the CHSS website. On 13 October 2026 the CHSS Center for Research and Quality listed its “UVR-SA Decision Tool (Critical Aortic Stenosis Calculator)” as “Currently under reconstruction”. The CHSS also hosts the Hickey 2007 PDF.
Model, equations and sources
Colan 2006 revised discriminant
Score = 10.98 x BSA [m2] + 0.56 x aortic annulus z + 5.89 x LAR
- 0.79 x (EFE grade 2 or 3: 1, else 0) - 6.78
Cutoff -0.65: score ≥ -0.65 favours survival with a biventricular circulation
Colan SD, McElhinney DB, Crawford EC, Keane JF, Lock JE. Validation and re-evaluation of a discriminant model predicting anatomic suitability for biventricular repair in neonates with aortic stenosis. J Am Coll Cardiol 2006;47(9):1858-1865. doi:10.1016/j.jacc.2006.02.020. Equation and cutoff verbatim from the PubMed abstract: 126 neonates; “With a cutoff of −0.65, outcome was predicted accurately in 90% of patients”; “among neonates with AS and a mitral valve area z-score >−2”. Minor (5%) measurement variation changed the predicted outcome in 9% (discriminant) and 15% (threshold system) of the validation cohort, which is why this page shows a band.
Direction of the cutoff: [partly unverified] the full text was not retrieved. ≥−0.65 is taken as favourable from the coefficient signs (larger structures raise the score) and from a 2025 comparison table listing “Discriminant ≥−0.65” as the biventricular cutoff (Wang et al., Clin Exp Pediatr 2025, doi:10.3345/cep.2025.02110).
Z scores: the model was built at Boston Children's with Boston z scores. Other references (PHN, Detroit) are not interchangeable in small left hearts (den Dekker et al., J Am Soc Echocardiogr 2013, doi:10.1016/j.echo.2013.07.022). The page withholds the score until you state the reference, and warns when it is not Boston.
Rhodes 1991 discriminant and threshold score
Score = 14.0 x BSA [m2] + 0.943 x aortic root index [cm/m2]
+ 4.78 x LAR + 0.157 x MV area index [cm2/m2] - 12.03
Score < -0.35 predicted death after a two-ventricle repair
Threshold score (1 point each; ≥2 unfavourable):
aortic root index ≤ 3.5 cm/m2, MV area index ≤ 4.75 cm2/m2,
LAR ≤ 0.8, LV mass index ≤ 35 g/m2
Rhodes LA, Colan SD, Perry SB, Jonas RA, Sanders SP. Predictors of survival in neonates with critical aortic stenosis. Circulation 1991;84(6):2325-2335. doi:10.1161/01.cir.84.6.2325. Equation, cut-off, units, measurement views and simplified (threshold) score read in the full text (Methods, Results, Tables 1 to 5). Cohort: 65 infants with aortic stenosis diagnosed in the first 2 months of life with LV dysfunction and heart failure; 43 were in the two-ventricle group (balloon or surgical valvotomy and/or coarctation repair, one with no intervention), of whom 14 died; age at first procedure (or at presentation) 0 to 120 days, BSA 0.15 to 0.35 m². Every measurement was divided by BSA; the paper does not state the BSA formula.
Performance in the derivation cohort: 88% correctly classified (4 survivors and 1 non-survivor misclassified), 91% of the 32 treated by 33 days; 7 of 8 in a later group of 8 (addendum). The threshold score: 2 or more points in 12 patients, all of whom died after two-ventricle repair; with 0 or 1 point, 2 of 31 died (the text gives 8%).
MV area formula. The paper calls it “the formula for an ellipse” but prints it as π × D1 × D2 / 2, which is twice the area of an ellipse with those diameters (π/4 × D1 × D2). Its own worked example (4.75 cm²/m² equals a 1.1 cm valve at BSA 0.2 m²) matches π/4, which this page uses. Because the paper cannot settle which formula produced its data, the details also show the score with the printed formula (it is higher by 0.157 × the MV area index); that value is not counted. The paper’s next worked example (a 3.5 cm/m² aortic root at BSA 0.2 m² given as 0.95 cm; it is 0.70 cm) has an arithmetic slip, so its examples point to π/4 but do not prove it. When the printed formula would put the Rhodes score on the other side of −0.35, the Rhodes vote is shown as uncertain and the bottom line becomes Borderline.
AVVI = left AV valve area / right AV valve area
All balanced CAVC: AVVI > 0.67
AVVI < 0.67 with a large VSD: consider a single-ventricle approach
Also shown: L / (L + R). AVVI 0.67 corresponds to L / (L + R) = 0.40.
Cohen MS, Jacobs ML, Weinberg PM, Rychik J. Morphometric analysis of unbalanced common atrioventricular canal using two-dimensional echocardiography. J Am Coll Cardiol 1996;28(4):1017-1023. doi:10.1016/s0735-1097(96)00262-8. Definitions and the 0.67 threshold verbatim from the PubMed abstract (103 patients).
[secondary source] Later CHSS work (Jegatheeswaran et al., 2010) used a modified index, left / total valve area, with ≤0.4 right-dominant and ≥0.6 left-dominant, as summarized by Ma 2019. The two indices use different denominators; check which one a paper uses before comparing numbers.
CHSS models (explained, not calculated)
CHSS 1 (Lofland 2001), form as given by two reviews:
Survival benefit = intercept + b1 (age at entry)
+ b2 (aortic valve z at the sinuses) + b3 (EFE grade)
+ b4 (ascending aorta diameter) + b5 (moderate or severe TR)
+ b6 (LV length z)
Positive favours Norwood; negative favours biventricular repair.
CHSS 2 (Hickey 2007, UVR-SA): predicted 5-year survival with
univentricular repair minus biventricular repair.
Positive favours univentricular; negative favours biventricular.
Lofland GK, McCrindle BW, Williams WG, Blackstone EH, Tchervenkov CI, Sittiwangkul R, Jonas RA. Critical aortic stenosis in the neonate: a multi-institutional study of management, outcomes, and risk factors. J Thorac Cardiovasc Surg 2001;121(1):10-27. doi:10.1067/mtc.2001.111207. Cohort, method and the direction of each factor are verbatim from the PubMed abstract.
[unverified] The CHSS 1 intercept, the six coefficients, the units of age and ascending aorta, the TR and EFE coding and the z-score reference. The full text (equation or nomogram) could not be retrieved; the publisher's site blocked automated access. The equation form and the sign convention above come from reviews that do not give the numbers: Ma XJ, Huang GY, Chin Med J 2019 (doi:10.1097/CM9.0000000000000375) and Mazza GA et al., Diagnostics 2024 (doi:10.3390/diagnostics14080823). A 2024 review table lists Daubeney 1999 as the z-score source for Lofland 2001 [secondary source]: Cantinotti M et al., Healthcare 2024 (doi:10.3390/healthcare12030348).
Hickey EJ, Caldarone CA, Blackstone EH, Lofland GK, Yeh T, Pizarro C, Tchervenkov CI, Pigula F, Overman DM, Jacobs ML, McCrindle BW, et al. Critical left ventricular outflow tract obstruction: the disproportionate impact of biventricular repair in borderline cases. J Thorac Cardiovasc Surg 2007;134(6):1429-1436. doi:10.1016/j.jtcvs.2007.07.052. Cohort, risk factors, R² and the discordant-management results are from the PubMed abstract. The sign convention, variable definitions, EFE grades and the statement about the CHSS website are from the copy hosted by the CHSS (Table 1 and its footnotes; Methods; Results).
[unverified] The UVR-SA coefficients. Table 1 lists an intercept and eight coefficients, but every minus sign was lost in the only copy that could be read. The scale of the output (fraction or percentage points) and the exact inverse transformation of the LVOT diameter could not be confirmed either, so the tool is not reproduced.
External performance (abstracts). Eicken A et al., Catheter Cardiovasc Interv 2010 (doi:10.1002/ccd.22363): 28 neonates, 2006 to 2008. Of 12 the UVR-SA assigned to a univentricular strategy, 2 had a successful biventricular circulation; of 16 it assigned to a biventricular strategy, 2 have a univentricular circulation. Sachdeva S et al., Cardiol Young 2020 (doi:10.1017/S1047951120002607): 68 neonates with intended biventricular repair, 1999 to 2017. No score's binary prediction was significantly associated with outcome. Among neonates predicted to need univentricular palliation, biventricular repair succeeded in 79.3% by CHSS 1 and 66.7% by CHSS 2. Mart CR, Eckhauser AW, Pediatr Cardiol 2014 (doi:10.1007/s00246-014-1009-0): in 20 neonates with hypoplastic left heart complex, a new score was more accurate than Rhodes, CHSS or the discriminant. No c-statistic for either CHSS model was found in what was retrieved. Hickey 2007 notes that the CHSS was enrolling a contemporary cohort to revalidate the model.
Bottom line: how the verdict is built
Lesion: critical AS / HLHS spectrum -> Rhodes 1991, Colan 2006
(both need: neonate with valvar AS = yes; Colan also MV area z > -2)
Lesion: unbalanced AVSD -> AV valve index (Cohen 1996)
Counted score = applies to the lesion AND every input entered and in range
Rhodes ≥ -0.35, Colan ≥ -0.65, AVVI ≥ 0.67 -> favours biventricular
Rhodes < -0.35, Colan < -0.65 -> favours single ventricle
AVVI < 0.67 with a large VSD -> favours single ventricle
AVVI < 0.67 without a large VSD; AVVI ≥ 1.5 -> not applicable
Uncertain = sensitivity band reaches both sides of the cut-off,
or (Rhodes) the paper's printed MV area formula would
put the score on the other side of the cut-off
No counted score -> no verdict
Any counted score uncertain -> Borderline (names the score)
All counted scores agree -> favours that route
Counted scores disagree -> Borderline
Scores built for the other lesion are listed as not applicable and never count. Rhodes and Colan are never combined with the AV valve index.
[interpretation] Cohen 1996 states that every balanced canal had AVVI > 0.67 and advises considering a single-ventricle approach when AVVI < 0.67 with a large VSD. Counting AVVI ≥ 0.67 as favouring biventricular repair, and AVVI < 0.67 without a large VSD as not applicable, is this page’s reading of that abstract. AVVI ≥ 1.5 (L/(L+R) ≥ 0.6, left-dominant by the modified index summarized by Ma 2019) is outside a threshold built for a small left ventricle.
[interpretation] Rhodes 1991 reports that a score below −0.35 predicted death after two-ventricle repair; its comparison with a first Norwood operation is for the threshold score (2 or more points: 6 of 13 survived a first Norwood, 0 of 12 a valvotomy). This page reads a score below −0.35 as favouring single-ventricle palliation. The Rhodes threshold score is shown in the details but not counted, because it comes from the same cohort and variables as the Rhodes discriminant.
Rhodes 1991 and Colan 2006 come from the same group; Colan 2006 validated Rhodes in 89 more neonates and re-derived the equation in all 126. Agreement between them is weaker evidence than agreement between independent models.
Published reproduction of the Rhodes equation: Minich LL et al., Am Heart J 1997 (doi:10.1016/s0002-8703(97)70152-2) report group means for 7 neonates with non-apex-forming left ventricles (BSA 0.2 m²; aortic root index 2.9 then 3.7 cm/m²; LAR 0.74 then 0.86; MV area index 2.3 then 3.2 cm²/m²) and mean Rhodes scores of −2.7 then −1.1. Because the score is linear, the score of the means equals the mean score: this page gives −2.60 and −1.13, within the rounding of the published means (±0.2). Rhodes 1991’s own Table 3 group means give −1.38 (deaths) and +0.70 (survivors), with the −0.35 cut-off at their midpoint. No published worked example or patient-level values were found for Colan 2006 or the AV valve index, and no official online calculator for Rhodes, Colan 2006 or the AV valve index was found (checked 13 October 2026: the Boston Children’s z-score calculator has none of them).
Sensitivity band
Each score is recomputed over every combination of each continuous input at −5%, 0 and +5% (measurements, BSA) and each z score at −0.5, 0 and +0.5 (Rhodes: 36 combinations; Colan: 34). The band is the lowest to highest result. If it reaches both sides of the cutoff the result is labelled borderline. Categorical inputs (EFE, population) are not varied.
Computed in this browser; nothing you enter in the calculator is stored or sent. These are reports of what published models say about their derivation cohorts. They were derived decades ago in single centres and perform less well in later cohorts; they are not a decision rule.