What six occupational health models have been measuring

Six occupational health models, built across different disciplines over five decades, converge on the same territory. Each was built within its own tradition, and between them they document the effects of these conditions in detail. What follows sets out what DFS reads in each.

Karasek’s Demand-Control model (1979)

Karasek’s central finding is that high demands combined with low decision latitude produce strain. Van der Doef and Maes (1999) confirmed this across sixty-three samples. In the Whitehall II study, low job control was the single largest contributor to the social gradient in coronary heart disease.

Decision latitude, as the model defines it, combines two things: the freedom to use one’s skills and the authority to make choices. In HDM organisations, what the instrument records as low decision latitude takes a specific form. Control is not restricted by automation or by procedure. It is restricted by choices that arrive already made, with no reasoning attached and nobody required to supply it. Asking is always possible; answering is owed by no one. So there is no route by which the lost control could be recovered. What the instrument records is a standing feature of the architecture, not a passing feature of the job. There is a further question the instrument does not ask: whether the person can see that the choice is going to fail. Where they can, low decision latitude means something sharper than having no say: perceiving the outcome in advance and having nowhere to take it.

Siegrist’s Effort-Reward Imbalance model (1996)

Siegrist’s framework is built around an unfair exchange: people give high effort and receive inadequate reward. Of the six models, his is the one that asks directly why people stay in conditions that harm them, identifying economic dependency and accumulated investment as the forces that make leaving difficult. These map closely to what DFS calls financial captivity and investment captivity.

The construct DFS identifies is a different one. Effort-reward concerns the fairness of exchange: input against output. What HDM creates is a person held accountable for outcomes they structurally cannot control. That is not an unfair exchange but an impossible demand.

The Whitehall studies (Marmot et al., 1978, 1991)

The Whitehall studies found that men in the lowest civil service grades had cardiovascular mortality rates more than three times higher than those at the top, and that the gap was not explained by standard risk factors. Measuring job strain at two points in time revealed cumulative effects on coronary heart disease risk that a single measurement missed entirely. Marmot’s interpretation points to autonomy and social participation as the explanatory factors.

DFS reads the same gradient through the mechanism that removes autonomy: choices that bind without their reasoning being answerable, the attribution of failure that follows, and the investment captivity that holds people in place long enough for the damage to compound.

The gradient runs on the first of Herman’s three conditions, the foreclosure of response. How often that foreclosure reaches a person depends in part on how much of their work is bound by choices made elsewhere, and that rises steadily on the way down; the cardiovascular gradient rises with it. The second condition, the individual attribution of failure, does not track grade the same way. How much of it lands on a person follows what the position is answerable for, and that is heaviest where authority over others sits together with directives received on the same terms. That is where Prins and colleagues found the depression, not at the bottom. The three conditions vary independently, so the harm does not fall along a single line down the hierarchy.

Maslach’s burnout model (Maslach and Leiter, 2016)

Maslach identifies six organisational mismatches — workload, control, reward, community, fairness, values — as the antecedents of burnout. The model is scoped to those antecedents and what they produce.

DFS proposes that burnout is the predictable midpoint of a single harm spectrum, and that investment captivity is what explains why exposure persists long enough for burnout and the clinical conditions that follow it to arrive. Captivity sits outside the model’s scope.

Hobfoll’s Conservation of Resources theory (1989)

Hobfoll’s loss spirals describe how initial resource loss leaves people less able to withstand further loss — of status, of autonomy, of professional self-concept. Each loss makes the next one harder to absorb. This is consistent with what sustained HDM exposure does over time: not a stable level of harm, but a deepening one.

DFS reads three things into the spiral. It is initiated by the architecture removing any means of affecting the outcome. It is kept moving by the individual attribution of failure, which turns each loss into evidence about the person. And it is held open by captivity, which keeps the person within reach of the next loss instead of out of it. The three are connected: what the spiral depletes — energy, confidence, standing — is what changing employer runs on, so each turn tightens the captivity that keeps the spiral running.

Warr’s Vitamin Model (1987)

Warr’s Vitamin Model proposes that job conditions affect wellbeing non-linearly: beyond a certain threshold, things get worse faster than a straight line would predict.

DFS does not derive where that threshold sits or why the curve bends where it does. That is an empirical finding about people. What DFS accounts for is the dose. The same configuration recurs because what a person perceives never reaches the choice that binds them, and captivity keeps them within range of the next round, so exposure accumulates instead of resetting between episodes.

What the pattern reveals

Across all six, one pattern holds. Each model identifies conditions associated with harm in detail; what produces those conditions sits outside the frame it was built to hold. None of the six locates the source of control removal. None treats the attribution of failure as a distinct and activating threat. Only Siegrist takes up why exposure persists. DFS proposes that all six are documenting downstream effects of a single upstream cause.

All six are conditional in form. They explain harm by what the work does to people — demands, control, reward, mismatch, resource loss, threshold. They do not explain it by what anyone is like.

Much of the workplace-harm literature beside them is attributional. It explains harm by what someone wants or does: hostile supervisors, bullies, dark-triad personalities. Both kinds describe harm that really happens. What separates them is the remedy each one leaves available. An attributional account ends in a person: the conduct is investigated, the person is moved or dismissed, and the case closes. The position is filled again, and whoever fills it finds the conditions the last occupant found. A conditional account ends in something that can be changed and stay changed. But only where the condition has been named. Nothing gets done about a condition nobody has written down.