Eighteen Months Stable: How Much Should the Injection Count
Everyone agrees his eighteen months on the injectable are real and sustained. What the team can't agree on is whether a structured risk assessment should treat that fact as new information, or as something the instrument is already counting somewhere else.
P.G., a 46-year-old man, has been under conditional forensic supervision for six years following a psychotic-episode assault that resulted in his index commitment; he has been on paliperidone palmitate, administered monthly and never missed, for the past eighteen months following two earlier relapses tied to oral non-adherence in the community. He works part time at a warehouse, lives independently in supervised housing, and has had zero symptomatic episodes, zero incidents of aggression, and zero missed clinical or probation appointments across the full eighteen months — a period substantially longer and cleaner than either of his two prior stable stretches before this admission. His treatment team is recommending a reduction in supervision intensity; the board reviewing that recommendation has asked for a formal structured violence risk assessment using the HCR-20 before ruling.
The HCR-20 will not hand the board a number to compare against a cutoff; it is a structured professional judgment instrument, and deliberately so — the evaluator rates each of the twenty factors for presence and for relevance to this particular man, then states a summary risk rating of low, moderate, or high that the ratings inform but do not compute. That design is exactly what makes the question here sharp rather than arithmetic. Its clinical and risk-management items already ask, separately, about his current insight, his response to treatment and supervision, and the feasibility of his ongoing plans — each of which his adherence arguably speaks to indirectly. The narrower question in front of the evaluator is whether his eighteen months of LAI adherence should ALSO be rated as its own distinct, risk-lowering factor on top of those related items, or whether doing so lets one real underlying change — his willingness and ability to stay in treatment — press on the summary rating from two or three directions at once under different labels. The evaluator's report will materially shape whether the board grants the reduced-supervision request, which is exactly why the rating question, and not just the ultimate recommendation, matters here.
Rating the instrument, before the board hearing
I want his adherence rated as a genuine, independent risk-lowering factor, not folded silently into the existing items. Chang et al.'s 2016 JAMA cohort of over 22,000 released prisoners found violent reoffending roughly forty percent lower during periods when antipsychotics were dispensed, hazard ratio 0.58, and Fazel et al.'s 2014 Lancet study of more than 82,000 patients found violent crime reduced about 45 percent during periods on antipsychotics, hazard ratio 0.55. Fazel reported a depot-specific figure too, 0.60 after adjusting for concurrent oral medication — which is his formulation, not an extrapolation from pills. And I'd stress the design, because it matters for what comes next: both were within-individual analyses, comparing each man to himself on medication and off it. An instrument that can't register a change of that size is missing real signal, not staying appropriately cautious.
You're right about the design, and I want to concede it properly rather than wave at the word "observational." Within-individual analysis does close the objection I would otherwise have made — that reliably adherent men simply differ from non-adherent men in insight, family support, and general functioning. Comparing each man to himself takes those stable differences off the table.
What it does not take off the table is the thing that varies inside one man's own life. Periods when someone is being dispensed medication tend to be the same periods when he is housed, in contact with services, and not using — none of which the within-individual design separates out from the drug. So I'll accept that antipsychotics carry a real effect, and still say the eighteen months in front of us is a bundle: the injection, and the housing, and the appointments he didn't miss. Rating adherence as its own separate risk-lowering factor, on top of already crediting him for good insight on the clinical items and a feasible plan on the risk-management ones, risks letting one bundled change lower the summary rating three times under three different names — and the manual's own revision went the other way, tightening the boundaries between items precisely to keep that from happening.
I don't think this hearing actually needs you two to resolve that rating question today, and I'd rather we not let the board's decision get held hostage to a genuine, unresolved methodological dispute. Whatever the right weight turns out to be on paper, continuing the injectable is an easy, low-burden recommendation regardless of who's right about the causal mechanism — he tolerates it, it's demonstrably working for him specifically, and recommending its continuation doesn't require us to have settled how much predictive credit it earns.
My proposal: the report states the association plainly, flags the double-counting concern explicitly rather than picking a side, and separates the RATING question — genuinely open — from the MANAGEMENT recommendation, which isn't.
Agreed: the report recommends continued LAI treatment and supports the reduced-supervision request, grounded in eighteen months of documented functional stability regardless of how the rating dispute resolves.
Not agreed, and stated as such directly in the report rather than resolved into one rating: whether adherence should be rated as an independent risk-lowering item or treated as already captured by the instrument's existing insight and plan-feasibility items. The psychologist's and psychiatrist's positions are both included as named, competing readings of the same eighteen months, with the risk manager's separate management recommendation standing regardless of which reading the board finds more persuasive.