Performance and Quality Management in an Oncology Hospital: From KPI to Management Decision

Most hospital dashboards answer a volume question: how many patients did we treat? In oncology that is rarely the question that matters. The real question is whether the right treatment reached the right patient at the right time, with the right technique and at an acceptable level of risk. This chapter takes that question as its starting point and builds a performance and quality system around a single idea: a key performance indicator (KPI) is only worth having if it leads to a management decision.

A KPI, in this view, is not a number that turns green or red. It is a tool that links a measurement to a target, a threshold, a deviation, a root cause, a comparison and, finally, an improvement decision. Quality indicators are useful when they are collected regularly and prospectively and fed back to clinicians and managers (Harden et al., 2022).

Not everything measurable is a good KPI

The most common beginner’s mistake is to turn every easily available data point into an indicator. A good KPI represents a clinically or managerially important process, has a reliable numerator and denominator, points to something the institution can act on, and is read alongside a comparison. Expert consensus processes show how selective this should be. A 2025 Delphi study for lung cancer radiotherapy started with 47 candidate indicators and finished with 29; the ESGO/ESTRO set for cervical cancer radiotherapy defined 19 indicators, each with what is measured, how and against which target. For every chosen KPI the chapter asks managers to write down its name, numerator, denominator, target, exclusions, data source and responsible owner: in other words, a KPI dictionary.

A four-layer KPI architecture

• Structure indicators show whether the capacity for good care exists: workload, infrastructure, machine downtime, training adequacy and access to radiotherapy (Salma et al., 2025).
• Process indicators show whether the patient moves through care on time and according to protocol: referral to first consultation, pathology result to treatment decision, appointment to first chemotherapy drug, simulation to treatment start.
• Outcome indicators measure what care brings to or costs the patient: survival, local control, toxicity, treatment completion, safety events and experience. They arrive late, which is why process indicators are needed as early signals.
• Balancing indicators check that improvement in one place is not causing harm in another. Every speed indicator should be paired with a safety balancer.

Waiting time is a clinical risk indicator

Waiting is often filed under patient satisfaction. The evidence says otherwise: in early-stage breast cancer, waiting 12 weeks or more from surgery to radiotherapy was associated with worse outcomes. The manager’s second task is to break total time into its parts (referral-consultation, consultation-simulation, simulation-plan, plan-approval, approval-treatment) so the bottleneck becomes visible. In one chemotherapy unit, design-oriented improvement cut the median morning treatment start from 83 to 45 minutes. But a shorter wait is never enough by itself; if pre-prepared regimens increase, the right patient-right regimen-right dose match must be monitored at the same time.

Reading surgical mortality correctly

Thirty-day surgical mortality is the classic outcome indicator, but its time window matters. Across 40,474 patients and eight cancer types, 30-day mortality ranged from 0.2% in breast surgery to 8.8% in gastric surgery. For complex procedures, 90-day mortality reflects surgical quality more legitimately, because many deaths occur after discharge. Composite indicators such as R0 resection rate, reoperation, readmission and “textbook outcome” should accompany it. The balancing risk is case selection: a hospital must not lower its mortality figure by quietly excluding high-risk patients who could benefit.

Radiotherapy quality assurance: the most direct link to harm

A 2025 meta-analysis of 13 randomised trials found that radiotherapy protocol deviations were associated with worse overall survival (hazard ratio 1.65) and worse progression-free survival (hazard ratio 1.79) (Jomy et al., 2025). Quality control here is not decorative compliance; it is a survival issue. Yet only 69% of phase III radiotherapy trials made quality assurance mandatory. The IAEA nasopharyngeal study shows what feedback can do: unacceptable plans fell from 64% to 11% after training and detailed quality assurance feedback.

Chemotherapy safety and incident reporting

In high-volume centres, even tiny error rates translate into real patients. An image-based preparation workflow cut the error rate from 0.045% to 0.017%. The chapter recommends monitoring risks caught before an error as well as detected errors: wrong drug, dose, patient or preparation, expiry and batch verification gaps, visual quality-control gaps.

On safety panels, a falling number of incident reports is not automatically good news. In one radiation oncology programme, a workflow change during COVID-19 reduced reports by about 25% (Volpini et al., 2022). The chapter asks boards to look at three numbers together: serious-harm incidents, near-misses caught without harm, and total reports. Rising near-miss reports with falling serious incidents may be the signature of a learning system.

Peer review and tumour boards

Peer review catches errors before treatment. Across 16,601 radiotherapy plans in a multinational network, the non-concordance rate was 4.5%, reaching 11% in head-and-neck cases. In thoracic SBRT, the target was for at least 90% of cases to pass peer review before 25% of the dose was delivered; completion rose from 19% to 79% as the process matured. The timing of quality control is itself an indicator: an error found after treatment teaches the system but cannot protect that patient.

Tumour boards are usually monitored by the proportion of patients discussed. That measures activity, not quality. In an Australian observation of 498 patients across 64 meetings, a decision was reached in 99% of cases, yet psychosocial factors were addressed at only 1.27 out of 5. MDT performance should therefore include availability of imaging and pathology information, discussion of comorbidities, and recording of patient preference.

Data infrastructure and patient-reported outcomes

Good KPIs cannot be produced from bad data. The hard part of a dashboard is not drawing charts but joining clinical, pathology, radiotherapy, medication and outcome data under one patient identity. Data quality (field completeness, timestamp accuracy, code standardisation, transfer delay) should itself be monitored.

The missing half of most outcome panels is what patients report. In a randomised trial of 766 patients, regular electronic reporting of 12 symptoms during chemotherapy was associated with better quality of life, fewer emergency visits and longer survival (median 31.2 versus 26.0 months) (Basch et al., 2017). The larger PRO-TECT trial found no survival difference but a longer time to first emergency visit (hazard ratio 0.84). PROMs therefore belong on the outcome panel and influence operational indicators too.

From KPI to decision: the five-stage cycle

• Define the standard, for example the share of high-risk SBRT cases that must pass peer review early.
• Detect deviation early using medians, percentiles, control charts and sub-unit breakdowns rather than monthly averages.
• Analyse the root cause with a process map before adding staff.
• Test the intervention with balancing indicators: if speed rises while errors or cancellations worsen, the process is not optimised.
• Feed back and standardise, through periodic comparative reports and learning across centres.

Alarms for the manager

Safety checks falling while speed rises; incident reports suddenly approaching zero; protocol deviations becoming normal; tumour board decisions that look complete but ignore the patient; and dashboards built on data whose source and calculation no one can explain.

The Turkish context

In Türkiye the framework rests on the Health Quality Standards (SKS) Hospital set, whose sixth version added chemotherapy, pathology, palliative care and indicator management sections, and on the Turkish Clinical Quality Programme, which includes colorectal and prostate cancer indicator cards and a case-level clinical quality score. International accreditations such as ESMO designation and OECI accreditation offer an additional quality basis. The monthly question for the manager is not “Are our KPIs green?” but “Which indicator points to a real risk or opportunity for patients, why has it changed, and what will we do about it next period?”

Source chapter: Kurutkan, M. N. (2026). Performance and quality management in an oncology hospital: From KPI to management decision (Chapter 2). Oncology Hospital Management Training Series, core text. Also available in Turkish.

References