Appropriate prescribing and waiting list governance: from variability to measurement

Prescriptive variability: a structural phenomenon

In the Italian context, the most recent data show that around 20% of specialist services provided do not generate a clinical value that can be documented against the patient's diagnostic-therapeutic pathway. This share, estimated by the Ministry of Health, is not a judgment on the quality of individual clinicians, but a structural feature of any complex healthcare system, where demand for services is mediated by multiple decision-making layers and the standardisation of prescribing criteria is inevitably incomplete.

Prescriptive variability is not an anomaly: it is an expected feature of any high-complexity healthcare system. The question is whether and how it is measured.

The Lazio Region has one of the highest prescription rates in the country: 1,570 prescriptions per 1,000 inhabitants, compared with a significantly lower national average and figures such as Veneto's, which stands at 890. This gap cannot be explained by differences in disease prevalence; it largely reflects prescribing patterns that the system does not yet have adequate tools to monitor systematically.

The first visit as a determinant of the care pathway

From a clinical and epidemiological standpoint, the first specialist visit plays a decisive role in how the care pathway subsequently unfolds. It is at that point that the initial diagnostic framework is set, further work-up is triggered, and the patient's care is structured.

The clinical framing established at the first visit has a determining effect on everything that follows: the services prescribed at that stage tend to shape the follow-up visits and instrumental exams that derive from them. In highly specialised settings, such as oncology centres or reference facilities for complex conditions, this dynamic is particularly significant: the prescribing threshold typical of those settings can trigger work-up pathways whose volume does not always match the patient's clinical profile at the time of access.

It follows that the appropriateness of the first visit, i.e. the consistency between the general practitioner's referral and the actual clinical indication, does not only affect the appropriateness of that single service, but the entire care trajectory that follows from it.

The appropriateness of the first visit does not only determine the appropriateness of a single visit: it determines the structure of the entire diagnostic-therapeutic pathway that follows.

The prescription: structured data and free text

From a technical standpoint, assessing appropriate prescribing runs into an intrinsic feature of the prescription document: a medical prescription contains structured information, such as patient identity, the name of the service, and the priority class, alongside a free-text field, the diagnostic question, which is unstructured and reports the clinical rationale for the prescription.

It is precisely in this field that the information most relevant to assessing appropriateness resides: the reason why that service was prescribed, with that priority, for that patient. The unstructured nature of the diagnostic question, subject to terminological, stylistic and level-of-detail variation between different clinicians, has historically made any systematic, large-scale analysis impossible.

The priority class system — U (urgent, within 24–72 hours), D (deferrable, within 30 days), P (schedulable, within 180 days), S (no indication of urgency) — offers a second level of assessment, complementary to clinical appropriateness. A priority assignment that is not consistent with the clinical picture produces capacity-saturation effects regardless of how pertinent the service itself is: an appropriate but over-prioritised service occupies short-term slots that might be needed for cases with genuine clinical urgency.

Clinical appropriateness and priority consistency are two distinct dimensions. Both affect the system's ability to guarantee timely access to those who need it.

Towards structured measurement: the regulatory framework and its operational implications

The Lazio Region Circular of 20 April 2026, issued as part of the National Waiting List Governance Plan 2025–2027 and Regional Council Resolution (DGR) 1344/2025, translated these elements into operational obligations for the General Management of SSR facilities: systematic oversight of priority-class consistency, realignment of the share of prescriptions in the "P – Schedulable" class to the national average, activation of Appropriate Prescribing Committees (CAPI), and the adoption of structured internal monitoring systems.

The implicit premise of these obligations is that measuring appropriateness is technically achievable at the level of a single facility. Until recently, this assumption was hard to sustain: the volume of reports and prescriptions produced daily by a hospital, combined with the unstructured nature of the diagnostic question, made any review that was not sample-based and retrospective impractical.

The evolution of natural language processing techniques has changed this picture. It is now possible to automatically extract clinically relevant information from the free text of reports and prescriptions, compare it against the guidelines of the leading scientific societies and against priority-assignment criteria, and produce a structured appropriateness assessment at the level of the individual service. ReportAId has developed and applied this capability to real caseloads in Italian hospitals, making it possible, for the first time in a systematic way, to measure a phenomenon that until now could only be estimated.

Analyses conducted in orthopaedics found that, in a significant share of cases, imaging services were prescribed ahead of the clinical pathway recommended by the relevant guidelines. In paediatrics, a significant proportion of antibiotic prescriptions showed characteristics not compliant with the indications of the leading scientific societies. These figures do not describe exceptions, but distributions to be expected in systems where prescribing standardisation is not yet supported by continuous monitoring tools.

The goal of measurement is not to assign responsibility, but to generate evidence on which to base training, organisational and health-policy interventions.

The shift from estimation to measurement is the necessary condition for any effective intervention on appropriate prescribing: without structured data at facility and pathway level, corrective actions inevitably remain generic. With that data, it becomes possible to identify the areas of greatest variability, design targeted interventions and evaluate their effectiveness over time.

Measurement as a support tool, not a control tool

Implementation of the Circular has opened a debate within the general-practice community. Organisations representing general practitioners have raised concerns about the prospect of systematic review of prescriptions by ASL medical directors, which part of the profession perceives as a surveillance mechanism that does not account for the complexity of the clinical relationship or the conditions under which primary care operates.

This is an understandable tension, and one worth addressing directly. Prescriptive variability is not the product of negligence or carelessness: it is the physiological result of a system in which continuous updating on guidelines is difficult, time available per patient is limited, and decision-support tools at the point of prescription are still insufficient. In this sense, prescriptive variability is first and foremost a system-level problem, not an individual one.

The relevant question is not who prescribed in a non-compliant way, but in which clinical and organisational contexts non-compliance concentrates, and which systemic interventions can reduce it.

The regional Circular itself explicitly states that the corrective actions expected from the CAPI Committees include training, clinical peer discussion and organisational adjustments, not sanctions. Interventions that produce lasting improvements in appropriateness are those that give clinicians structured feedback and support tools.

From this perspective, the availability of structured appropriateness data at facility and clinical-area level is not a surveillance tool aimed at individual physicians: it is the information base needed to design targeted training pathways, update local protocols based on emerging issues, and evaluate the effectiveness of interventions over time. Its value is greatest when it is oriented towards shared improvement rather than the attribution of blame.