Overall Equipment Effectiveness (OEE) has become the de facto performance measure through which Total Productive Maintenance (TPM) programmes are justified, monitored and audited. Yet the metric and the programme were both formalised for discrete, repetitive assembly environments, whereas the process industries — cement, chemicals, pharmaceuticals, pulp and paper, sugar, food and beverage — operate coupled, continuous or semi-continuous plants in which availability is constrained by campaign scheduling, cleaning and changeover validation, thermal inertia and upstream–downstream buffering. This review addresses the resulting mismatch. Its objective is to establish, through a critical and comparative synthesis rather than a descriptive summary, how TPM-to-OEE research has evolved, which methodologies and data sources dominate, what is robustly established, and what remains unresolved for continuous-process settings. A structured search of Scopus-, Web of Science- and PubMed-indexed sources was executed over the period 2021–2026, supplemented by seminal earlier work, and the retrieved corpus was organised around four themes: metric formalisation and its variants, single-site diagnostic case studies, structured design and deployment frameworks, and the digital, statistical and sustainability extensions of TPM. Two peer-reviewed Elsevier studies are adopted as base papers and dissected in depth: a Design for Lean Six Sigma (DMADV) TPM and reliability framework deployed in an Active Pharmaceutical Ingredient plant, and a lean-tools productivity study in a cement plant. The synthesis yields three principal findings. First, reported OEE gains are almost entirely attributable to availability recovery, while performance-rate losses — the dominant loss class in continuous plants — remain under-diagnosed. Second, the evidential base is overwhelmingly single-site, pre–post and uncontrolled, so effect sizes cannot be separated from concurrent demand, feedstock and seasonal variation. Third, no published framework links individual TPM pillars quantitatively to specific OEE components in a process-industry setting, which leaves pillar sequencing a matter of practitioner judgement rather than evidence. Two defensible gaps are formalised from these findings — a methodological–causal gap and a process-industry contextualisation and generalisability gap — and a conceptual research framework is proposed that couples loss-stratified OEE measurement with reliability statistics and a pillar-to-component attribution model, evaluated through a controlled multi-period, multi-line design. The review distinguishes throughout between what the literature has established and what is advanced here as a recommendation.