Garden Classification of Femoral Neck Fracture

Stage an intracapsular femoral neck fracture I to IV from completeness and displacement — and see why the four grades agree so poorly between observers that most practice collapses them to two.

Last updated 8 Oct 2026

Garden stage from completeness and displacement

Views + completeness + displacement

Imaging availableGarden’s stages are defined on the fracture line and the position of the head, and both need an anteroposterior and a true lateral view: an undisplaced-looking fracture on the AP alone can be posteriorly tilted on the lateral, which is the difference between stage II and stage III. Imaging modality also changes the agreement figures, and by a lot. In a series of 886 elderly patients read by eight observers, four-type agreement on radiographs ran from kappa 0.18 to 0.43, while on CT the senior readers reached 0.56 to 0.76 — and the junior readers did not improve at all (0.28 to 0.35).

Is the fracture line complete across the femoral neck?This is the stage I / stage II distinction and it is the one that falls apart in practice. An incomplete fracture is one in which part of the neck is still in continuity, which in older adults usually means the valgus-impacted pattern where the head is tilted into valgus on the intact inferior cortex. The 886-patient series was written specifically to ask whether an incomplete femoral neck fracture exists at all in older adults, and found that collapsing I and II into one group raised agreement to kappa 0.76 to 0.90.

Displacement of the head relative to the neckJudged on both views together, from the trabecular pattern of the head against that of the neck and the acetabulum. Incompletely displaced means the fragments have moved but retain contact — the classic stage III appearance is the head rotated so that its trabeculae no longer line up with the acetabulum. Completely displaced means no contact and the head free in the acetabulum. A numerical analysis of 200 fractures measured the articulo-trochanteric distance and the anteroposterior head position and found stages I and II indistinguishable from each other, and stages III and IV indistinguishable from each other.

Garden stage III — complete fracture, incompletely displacedExample

Anteroposterior and true lateral radiographs; the fracture line crosses the whole neck; the head has moved but retains contact with the neck

The four stages, as Garden defined them

stage I = undisplaced incomplete (including valgus-impacted) · stage II = undisplaced complete · stage III = complete, incompletely displaced · stage IV = complete, completely displaced

two questions, not four grades: the whole classification is the answers to “is the line complete?” and “has the head moved, and how far?”. Everything else about it follows from those two
the split that actually survives: I and II are undisplaced, III and IV are displaced. That boundary is reproducible; the four-way division is not. Ten observers on 100 fractures got kappa 0.31 for four groups and 0.52 for two; eight observers on 886 patients got 0.18 to 0.43 for four types and 0.76 to 0.90 for three
the measurement argument, separately: a 1988 analysis of 200 fractures measured the articulo-trochanteric distance and the anteroposterior head position against 100 normal hips and found I indistinguishable from II and III indistinguishable from IV. So the four grades collapse to two not only because observers disagree but because the underlying displacement does not come in four sizes
nothing here is ordinal: the stages are names for radiographic appearances. There is no sum, no interval and no arithmetic — a mean Garden stage is not a quantity
derivation: Garden RS, “Low angle fixation in fractures of the femoral neck”, J Bone Joint Surg Br 1961;43:647–663. The paper was not reachable for this page; the stage definitions here are quoted from Radiopaedia and from the University of Washington trauma radiology reference resource, both of which reproduce them in the same words
what this page does not do: it does not estimate fracture risk, and no page on this site does. FRAX is the instrument most readers are after and it is licensed: copyright Osteoporosis Research Ltd, UK, a registered trademark, with an unpublished algorithm and no stated terms for third-party use

Worked example

Anteroposterior and true lateral radiographs; the fracture line crosses the whole neck; the head has moved but retains contact with the neck
Both views are available, so the first rule — which refuses to stage from an anteroposterior film alone — does not fire
The line is complete, so this is not stage I
The head has moved, so this is not stage II
Contact is retained, so it is stage III and not stage IV
Now change one entry. If the head had lost contact altogether the answer would be stage IV — and the 1988 measurement study found no measurable difference in displacement between III and IV, so the two are one group in all but name
Change the imaging instead, to the anteroposterior film alone, and the page refuses to stage at all rather than reading the lateral plane off a view that does not show it

Published agreement for the Garden classification

Fleiss’ kappa between observers and Cohen’s kappa within, read against Landis and Koch in the 2022 series. Two things to read off this table. The four-grade figures span 0.18 to 0.76 — more than fourfold — and the low end is poorer agreement than chance-corrected guessing would embarrass. And collapsing I and II moves agreement further than changing modality, buying experience or reading twice: every collapsed figure is 0.52 or above and most are above 0.76.
Study	Observers and images	Four grades	Collapsed groups
Van Embden et al 2012	10 trauma surgeons and residents, 100 intracapsular fractures	kappa 0.31 (inter-observer)	kappa 0.52, displaced vs non-displaced
BMC Surgery 2022, radiographs	8 observers (2 professors and 2 residents each from radiology and orthopaedics), 886 elderly patients	0.24 to 0.43 among professors, 0.18 to 0.35 among residents	0.76 to 0.90 for three types (I+II, III, IV)
BMC Surgery 2022, same observers on CT	as above	0.56 to 0.76 among professors, 0.28 to 0.35 among residents	0.76 to 0.90, unchanged by modality
BMC Surgery 2022, intraobserver, radiographs	as above	0.38 to 0.42 among professors, 0.26 to 0.36 among residents	not reported separately
Frandsen et al 1986, as cited by Eliasson 1988	not stated in the citing paper	“considerable” interobserver variation, no figure given	—

What the two displaced grades are reported to carry

Every figure here is a property of the series it came from. The honest summary is that avascular necrosis is commoner after displaced fractures than undisplaced ones, that the published rates span 5 to 40%, and that no series read for this page is large enough to put a number on a particular Garden grade with confidence.
Figure	Source and cohort	The caveat
Avascular necrosis about 7% (stage I), about 37% (stage III)	University of Washington trauma radiology reference resource, attributed there to a 2015 Radiographics review	No cohort is given for either number in the source read, and none is invented here
Avascular necrosis 6% undisplaced vs 23% displaced; nonunion 18% vs 38%	A retrospective series of 30 fractures in 28 patients, at least two years of follow-up	Thirty fractures. A single event moves any of these percentages by three points or more
Avascular necrosis 4% overall; no association with displacement	56 patients (42 displaced, 14 undisplaced) after cannulated screw fixation, 2024	p ≥ 0.999 for the displacement association — a null result in a small sample, not a demonstration of no effect
“reported rates ranging from 5 to 40%”	The same 2024 paper’s summary of five prior studies	An eightfold spread. Treatment, follow-up length and the definition of necrosis all differ between the series

Two questions, four names, and the two that survive contact with a second reader

Garden’s classification asks two things about an intracapsular femoral neck fracture: whether the fracture line crosses the whole neck, and whether the head has moved. Stage I is undisplaced and incomplete, which in older adults usually means the valgus-impacted pattern. Stage II is complete and undisplaced. Stage III is complete and incompletely displaced, with contact retained. Stage IV is complete and completely displaced. Those four sentences are the whole system, and they have been reproduced in textbooks since 1961 because they are clear and because they divide the injury along the line that matters most to the blood supply of the head.

The trouble is that observers do not reproduce them. Ten trauma surgeons and residents reading 100 intracapsular fractures reached an inter-observer kappa of 0.31 for the four grades. Eight observers reading 886 elderly patients reached 0.18 to 0.43 on radiographs, and the residents among them barely improved on CT, from 0.18-0.35 to 0.28-0.35. What changes the picture is collapsing the grades: merge stages I and II into one undisplaced group and agreement jumps to 0.52 in the first study and 0.76 to 0.90 in the second. That is not a statistical trick. A 1988 study measured displacement in 200 femoral neck fractures and found stage I indistinguishable from stage II and stage III indistinguishable from stage IV, and concluded that two groups were probably sufficient. The four grades do not merely get confused; two of the four boundaries describe differences that are not reliably there.

So the practical content of a Garden stage is the displaced/undisplaced distinction, and this page says so in every answer. It also refuses to assign a stage from an anteroposterior film alone, because the plane that separates stage II from stage III is usually the lateral, and the same 1988 measurement study found posterior tilt of the head present even in the undisplaced grades.

One thing this page deliberately does not do is estimate fracture risk. A femoral neck fracture in an older adult is a fragility fracture, and the skeletal work-up belongs with the bone and mineral records — the T-score and Z-score calculator, the bone turnover markers, corrected calcium and vitamin D status. The risk instrument most readers want is FRAX, and it is not here and is not approximated: its copyright is held by Osteoporosis Research Ltd, its name is a registered trademark, and its coefficients have never been published, so any implementation would be a guess wearing a licensed name.

A classification is not a score. Garden IV is not one unit worse than Garden III, a Salter-Harris IV is not twice a II, and nothing on this page adds up — these are named categories, not an ordinal measurement, and a mean or a difference taken over them has no meaning. This page states which published category the entered findings fall in and what the literature reports for that category in the cohorts it was measured in. It renders no clinical or surgical decision: not whether to operate, not which implant, not whether to take or withhold an image, and not whether a limb can take weight. Those are the treating clinician’s, and the classification is one input among many. Agreement between observers is part of what a classification means. A category two experienced readers assign differently on the same film is not a measurement of the fracture, and the published kappa values for these systems run from almost nothing to almost perfect. Each page here prints them with the number of observers and the number of radiographs they came from.

Frequently asked questions

What are the four Garden stages?

Stage I is an undisplaced incomplete fracture, including the valgus-impacted pattern. Stage II is a complete but undisplaced fracture. Stage III is a complete fracture that is incompletely displaced, with contact retained between the fragments. Stage IV is a complete fracture that is completely displaced. The wording here follows Radiopaedia’s and the University of Washington trauma radiology resource’s reproductions of Garden’s 1961 paper.

Why do so many surgeons use only displaced and undisplaced?

Because that is the division the evidence supports. The four-grade inter-observer kappa was 0.31 across ten observers and 100 fractures, and 0.18 to 0.43 across eight observers and 886 patients; collapsing stages I and II raised those to 0.52 and to 0.76–0.90. Separately, a measurement study of 200 fractures found I and II indistinguishable and III and IV indistinguishable, and concluded two groups were probably enough.

Is a Garden IV twice as bad as a Garden II?

No, and the question has no answer, because the stages are not a measurement. They are four names for four radiographic appearances. Nothing adds, nothing subtracts, and a mean Garden stage across a series of patients is not a quantity. The only comparison the stages support is displaced against undisplaced.

Does CT improve the staging?

For experienced readers, in one series. Among eight observers reading 886 elderly patients, the two radiology and two orthopaedic professors improved from 0.24–0.43 on radiographs to 0.56–0.76 on CT, while the four residents stayed at 0.28–0.35. The collapsed three-group figure was 0.76–0.90 whichever modality was used, so CT closes part of the gap between the four grades and does nothing for the division that is already reliable.

Why is there no FRAX calculator on this site?

Because FRAX is licensed and its algorithm is unpublished. The copyright notice reads “©2008-2026 – Osteoporosis Research Ltd, UK”, FRAX and FRAXplus are registered trademarks, and neither the calculation tool nor the about page states terms permitting a third-party implementation. The model coefficients are not in the public domain, so a calculator here could only be an approximation of an algorithm nobody outside Sheffield can see — which would be a different instrument using a licensed name.

Can this page tell me whether the hip needs replacing?

No. It states which published stage the entered findings fall in and what the literature reports for that stage in the cohorts it was measured in. The operation, the implant and the timing are decisions for the treating team, and the classification is one input among many — age, function before the injury, bone quality and comorbidity all bear on them and none of them is on this page.

Related calculators

References

Garden RS. Low angle fixation in fractures of the femoral neck. J Bone Joint Surg Br. 1961;43:647–663. Citation read in the reference list of the University of Washington trauma radiology reference resource (below); the paper itself was not reachable, so every criterion below is quoted from a named reproduction rather than from Garden.
Radiopaedia. Garden classification of subcapital femoral neck fractures. The four stages verbatim: “Garden stage I: undisplaced incomplete, including valgus-impacted fractures”; “Garden stage II: undisplaced complete”; “Garden stage III: complete fracture, incompletely displaced”; “Garden stage IV: complete fracture, completely displaced”. Also reports “poor intraobserver reliability” for the classification.
University of Washington, Trauma Radiology Reference Resource. Garden classification of femoral neck fractures. faculty.washington.edu/jeff8rob. Type 1 “Incomplete, nondisplaced, including valgus impacted fractures” with “~7% AVN”; type 2 “Complete, nondisplaced”; type 3 “Complete, incompletely displaced” with “~ 37% AVN”; type 4 “Complete, completely displaced”. The two AVN figures are attributed there to Sheehan SE et al, Radiographics 2015;35(5):1563–1584, which was not itself read, and no cohort is given for either — stated on the page as a figure without a cohort rather than presented as one.
Does Garden type I incomplete femoral neck fracture really exist in older adults? To evaluate the stability and consistency of Garden classification. BMC Surg. 2022;22:276. X-ray and CT images from 886 elderly patients, read by 8 observers (two professors and two residents from each of radiology and orthopaedics). Four-type Garden on radiographs: interobserver kappa “ranged from 0.24 to 0.43” among professors and “from 0.18 to 0.35” among residents; intraobserver “ranged from 0.38 to 0.42” and “from 0.26 to 0.36” respectively. On CT, “professors consistency (0.56 to 0.76) was higher than residents (0.28 to 0.35)”. Collapsed to three types (I+II, III, IV) the values “ranged from 0.76 to 0.90”. Fleiss’ kappa between observers, Cohen’s within, read against Landis and Koch.
Van Embden D, Rhemrev SJ, Genelin F, Meylaerts SAG, Roukema GR. The reliability of a simplified Garden classification for intracapsular hip fractures. Orthop Traumatol Surg Res. 2012;98(4):405–408. “Ten observers, trauma surgeons and residents, from two different institutes classified 100 intracapsular femoral neck fractures.” “The inter-observer kappa for the Garden classification was 0.31”; “An agreement of κ0.52 was observed if the Garden classification was simplified” to displaced versus non-displaced. Abstract read at em-consulte.com/article/726322; the full text was not reachable.
Eliasson P, Hansson LI, Karrholm J. Displacement in femoral neck fractures: a numerical analysis of 200 fractures. Acta Orthop Scand. 1988;59(4):361–364. Articulo-trochanteric distance and anteroposterior head position measured in 200 femoral neck fractures against 100 normal hips. Stages I and II were similar to each other, stages III and IV were similar to each other and more posterodistally displaced, and the authors conclude that two groups are probably sufficient. Also records, citing Frandsen et al 1986, that “The interobserver variation in the assessment of fracture staging according to Garden is considerable”.
Scattergood SD, Berry AL, Flannery O, et al. Low avascular necrosis rates in femoral neck fractures: efficacy of cannulated screw fixation. Eur J Orthop Surg Traumatol. 2024. 56 patients (42 displaced, 14 undisplaced); “Two (4%) patients developed AVN”; “no correlation between AVN risk or reoperation and displacement” (p ≥ 0.999); and published rates cited as “reported rates ranging from 5 to 40%”.
Abstract of a retrospective series of 30 femoral neck fractures in 28 patients with at least two years of follow-up, read at orthobullets.com/evidence/15727403; the page did not display the title, authors or journal, so the citation cannot be given. Avascular necrosis and nonunion “were 6% and 18%, respectively, among patients with undisplaced (Garden stages 1 and 2) fractures and 23% and 38% among those with displaced (Garden stages 3 and 4) fractures”. Thirty fractures is a very small denominator and the page says so wherever these numbers appear.
FRAXplus. Calculation tool and About pages, fraxplus.org, read 9 October 2026. “©2008-2026 – Osteoporosis Research Ltd, UK”; “FRAX® and FRAXplus® are registered trademarks”; the tool “was launched in 2008 by a team based within the World Health Organisation (WHO) Collaborating Centre” at Sheffield and “was neither developed or endorsed by WHO”. Neither page states any terms permitting third-party implementation, and the model coefficients are not published. That is why FRAX is named on this page and not built.

Not medical advice. For healthcare professionals and education. Reference intervals vary by laboratory and assay — always use your own laboratory's. Never base a dose or a treatment decision on this page alone. Full disclaimer at calcengines.com/disclaimer/