14An Evidence-Based Prosthesis Selection Algorithm

FWhy an algorithm, not a reflex

Faced with a broken ossicular chain, the inexperienced surgeon reaches for a favourite prosthesis; the experienced one runs a sequence of decisions. The difference matters because the choice of construct is not a single judgement but a small cascade: first the geometry the reconstruction must reproduce, then the environmentit must survive in, and only then the device, its protection and the timing of surgery. Each step has a defensible evidence base, and laying them out as an explicit algorithm makes the reasoning teachable, reproducible and — when a result disappoints — auditable.

The crucial conceptual move is to separate the two questions the chain poses. Anatomy chooses the construct. Which ossicles survive, and whether the stapes footplate is mobile, determines whether a partial or total prosthesis is even possible and where it must seat. The environment predicts the result. Disease activity, mucosal health, drainage and prior surgery govern how well any construct will work, far more than the brand or material chosen. Confusing these two questions is the commonest source of muddled decision-making; keeping them distinct is the backbone of the algorithm that follows [1971, 2001].

FNode one: the surviving ossicles choose the geometry

The first node is purely anatomical and is resolved by inspection in the oval window niche. It runs as three nested questions, in order. Is the stapes footplate mobile?If it is fixed or absent, neither a partial ossicular replacement prosthesis (PORP) nor a total ossicular replacement prosthesis (TORP) will transmit, and the problem has moved to the footplate — a stapes procedure or alternative strategy must come first. Is the stapes superstructure intact? If the crura and capitulum survive and offer a stable seat, a shorter PORP rests on the stapes head; if they are eroded away, only a longer TORP reaching the footplate will bridge the gap. Is the malleus handle usable? Couple to it whenever it is present, because a malleus-coupled construct retains part of the native lever and is more stable than one terminating only on the drum.

This is the practical core of Austin’s classification, which sorts ossicular defects precisely by the presence of the malleus handle and the stapes superstructure, because those two features dictate the columellar geometry a reconstruction must reproduce [1971]. Kartush extended the matrix to capture an intact chain, malleus-head fixation and stapes fixation, and singled out the capitulum-to-malleus strut as the most stable construct — the geometry the algorithm steers toward whenever the anatomy allows [1994]. Black’s analysis of 535 ossiculoplasties reached the same conclusion from the outcome side, naming the malleus the single most important ossicle for prosthesis stability and long-term gain [1992]. Walk the three questions on the explorer below and watch one construct survive.

Notice what this node does notdecide. It does not choose a material, a length in millimetres, or whether to operate today or in six months. It narrows a confusing market down to a single geometric category — PORP onto the stapes head, TORP onto the footplate, or “fix the footplate first” — and leaves every other choice to the nodes that follow. Resolving geometry first is what keeps the rest of the algorithm orderly.

TNode two: grading the ear environment

Once the geometry is fixed, the second node grades the bed the construct must live in. Decades of outcome data converge on the same uncomfortable truth: the middle-ear environment outweighs the prosthesis. Dornhoffer’s multivariate analysis of 200 ears built the OOPS staging index from mucosal status, ossicular remnant, ossiculoplasty type, drainage and prior surgery, and repeatedly showed these factors — not the material chosen — predict the achievable air-bone gap [2001]. The Middle Ear Risk Index (MERI) embeds ossicular status inside a wider weighted score that adds otorrhoea, cholesteatoma, prior surgery and even smoking [1994], while Black’s SPITE method groups twelve preoperative features as Surgical, Prosthetic, Infection, Tissue and Eustachian factors [1992]. Bellucci’s older otorrhoea grading is the simplest of all, tying the wet ear directly to eustachian and mucosal failure and flagging the ear that should not be reconstructed today [1973].

These scores overlap heavily, as the matrix below shows: ossicular status, drainage, mucosal health and prior surgery recur in nearly all of them because they are the genuine drivers of outcome. The most recent and largest addition, the multi-centre Ear Environment Risk (EER) scale derived from 1679 ossiculoplasties, quantifies how revision surgery, a canal-wall-down cavity, active drainage and a lateralised or blunted drum each widen the postoperative gap [2025]. Select each score and see which factors it weighs.

A vital caveat keeps this node honest. When MERI, SPITE and OOPS were compared head-to-head in the same cohort, all correlated only weakly with hearing outcome, SPITE was the only independent predictor of the postoperative gap, and none was clearly superior [2020]. The lesson is not to abandon risk assessment — the environment plainly dominates — but to treat any single number as a structured promptto consider the right factors, not a deterministic verdict. The score informs counselling, protection and timing; it does not override anatomy or the surgeon’s eyes.

TWhat each branch predicts

Because anatomy chooses the branch and the branch carries its own prognosis, the algorithm lets you set expectations the moment the construct is decided. The headline comes from a meta-analysis of 40 studies and 4,311 ears: a PORP is significantly more effective than a TORPat chain reconstruction, with a combined risk ratio of 1.28 (95% CI 1.17–1.41) favouring the partial prosthesis and better long-term stability [2013]. The chart below renders that relative effectiveness.

Read the figure correctly. It does notmean a TORP is a poor device or that you may choose a PORP when the superstructure is gone — you cannot, because the PORP would have nothing to sit on. It means the defecta TORP is summoned to repair is inherently harder: the prosthesis is longer, balanced on a small mobile footplate rather than nestled in the capitulum, and the absent superstructure usually signals more aggressive disease that has already emptied the niche. The branch’s lower average is therefore a property of the earas much as of the device. The same logic links the two nodes: a worse anatomical branch tends to coincide with a worse environmental score, and the two together — not either alone — set the realistic ceiling on hearing [2013, 2001].

CTuning the plan: protection, timing, material

With geometry chosen and environment graded, the final node tunes the plan. Three decisions hang on the environmental score rather than on the anatomy.

• Cartilage protection.Wherever a rigid synthetic head abuts a thin drum — and especially in a high-risk, poorly aerated or lateralised ear — interpose a cartilage shield to curb the slow extrusion that plagues uncapped heads. This is near-default for alloplastic PORP and TORP heads [2025].
• Primary versus staged.The wet ear and the doubtful mucosa argue for staging: Bellucci’s higher otorrhoea grades and a frankly draining cavity at surgery should push you to defer reconstruction until the ear is dry and aerated, rather than seat a prosthesis in a hostile bed [1973, 2001].
• Material. This is the last and least decisive choice. In favourable ears a sculpted autograft, titanium, hydroxyapatite or a hybrid all perform comparably, because the environment dominates; reach for an autograft when a healthy, disease-free ossicle is available, and for a consistent off-the-shelf alloplast when none is, or when standard geometry is needed [2001].

Each of these is a downstream adjustment, not a re-opening of the geometry decision. The discipline of the algorithm is that questions are answered once, in order, and lower nodes never overturn higher ones: a bad environmental score makes you protect, stage and counsel harder, but it does not turn an absent superstructure into a PORP.

CThe algorithm at the microscope

Distilled to what happens in the few seconds after the chain is inspected, the algorithm reads:

The mature view is neither dogmatic nor casual. The anatomical nodes are answered first and firmly, because they determine what is mechanically possible [1971, 1994]. The environmental node is answered with humility, because the scores that grade it predict only weakly and the ear’s biology will have the final word [2020, 2001]. Run the sequence the same way every time, let anatomy choose the construct and the environment set the expectation, and you will make a choice you can defend long after the operation — whatever the audiogram eventually shows [2013, 2025].