Continual improvement, integrated systems and a capstone gap assessment

Two years after Rachel Okoye's first slide, Harrowgate Foods Ltd. had a working safety management system. It had a policy that workers recognized, a risk register that reflected the floor, critical controls that were verified, contractors who were prequalified, emergency plans that had been tested on nights, an internal audit program with worker auditors, and external recognition. It also had a list of things that were not yet good enough. That list is the subject of this final lesson: the act stage of plan-do-check-act, where an organization deals with what went wrong, improves what works, and keeps the system from going stale.

The lesson closes with a capstone: a gap assessment that brings together everything in the course, applied to a site that has not yet been through the process.

Incidents, nonconformities and corrective action

Management system standards treat incidents and nonconformities through one process. An incident is an occurrence arising out of or in the course of work that could or does result in injury or ill health. A nonconformity is the non-fulfilment of a requirement, whether a legal requirement, a requirement of the standard or one of the organization's own. The organization must react in a timely way, evaluate with the participation of workers whether corrective action is needed to eliminate the root causes, implement the action, review risk assessments where needed, review the effectiveness of the action, and make changes to the system if necessary. Documented information must be kept as evidence.

HSE-403 covered incident investigation in depth. Here the focus is on what the management system does with the results. Every investigation and every audit nonconformity at Harrowgate Foods Ltd. fed one corrective action register, with each action recorded against its source, its cause, its owner, its due date and its verification. The register made patterns visible: when three separate actions from three different investigations all pointed to the same weak handover between day and night maintenance, it was clear that a system change was needed, not three local fixes.

Corrective action that works

Corrective actions fail in predictable ways. They target the person rather than the conditions, such as retraining the worker who was hurt. They rely on the weakest controls, such as a new sign or a memo. They are closed when the work order is issued rather than when the fix is in place. Or they are never checked for effectiveness, so nobody knows whether they worked. Harrowgate Foods Ltd. adopted four rules for its register.

• Address causes, not symptoms. Each action had to name the cause it addressed, drawn from the investigation or audit finding.
• Prefer stronger controls. Actions were tagged by level of the hierarchy of controls, and reviewers challenged any serious finding answered only with training or a procedure change.
• Close on evidence. An action was closed only when the person who raised it, or a committee member, had seen the fix in place.
• Check effectiveness. For high-risk actions, a follow-up check three to six months later confirmed whether the problem had come back.

# Corrective action register summary: closure and effectiveness.
actions = [
    # (id, level of control, closed, effectiveness verified, recurred)
    ("CA-031", "engineering",    True,  True,  False),
    ("CA-032", "administrative", True,  True,  True),
    ("CA-033", "elimination",    True,  True,  False),
    ("CA-034", "training",       True,  False, False),
    ("CA-035", "engineering",    False, False, False),
    ("CA-036", "administrative", True,  True,  True),
]
closed = [a for a in actions if a[2]]
verified = [a for a in closed if a[3]]
recurred = [a for a in verified if a[4]]
print(f"Closed: {len(closed)} of {len(actions)}")
print(f"Effectiveness verified: {len(verified)} of {len(closed)} closed")
print("Recurred after verification:", ", ".join(f"{a[0]} ({a[1]})" for a in recurred))

Closed: 5 of 6
Effectiveness verified: 4 of 5 closed
Recurred after verification: CA-032 (administrative), CA-036 (administrative)

The summary told its own story. The two actions where the problem came back were both administrative controls. That fed the next management review, which asked for the two problems to be reopened with stronger controls considered first.

Continual improvement

Continual improvement means recurring activity to enhance performance. It does not mean improving everything at once, or that every indicator must improve every month. Management system standards expect the organization to improve the suitability, adequacy and effectiveness of the system by enhancing OHS performance, promoting a culture that supports the system, promoting worker participation in improvement actions, communicating the results, and keeping documented information as evidence.

Improvement opportunities come from everywhere in the system: monitoring trends, audits, management review, worker suggestions, incident investigations, changes in technology, new knowledge about hazards, benchmarking with other companies and changes in standards. Harrowgate Foods Ltd. kept an improvement log separate from the corrective action register, because improvements are not only fixes for failures. Ideas from the floor, such as the sanitation crew's group lockout box or the forklift operators' proposal for one-way aisles, went into the log, were assessed, and if adopted were credited to the people who proposed them.

Keeping the system from going stale

Mature systems face a different risk from new ones. In the first year, everything is new and attention is high. By the third year, procedures can become routine, audits can become predictable and management review can become a ritual. Elena watched for signs of drift: audits that found nothing, hazard reporting that fell without a clear reason, the same names on every improvement team, actions closed without evidence, and management reviews that took less time each year. She changed the internal audit focus each year, rotated auditors between sites, invited a peer from another food company to audit one process, and asked the joint occupational health and safety committees each year what the system was no longer catching.

Integrating with quality and environmental systems

Harrowgate Foods Ltd. already ran a certified food safety system, and Aisha Rahman's team was considering ISO 9001 for quality and ISO 14001 for environmental management. All three standards share a common high-level structure: the same clause numbers and titles for context, leadership, planning, support, operation, performance evaluation and improvement, and much shared text. ISO 9001 and ISO 14001 were both revised in 2026, and ISO 45001 is under revision, with a new edition expected in 2027. The shared structure carries through these revisions, which makes integration practical.

An integrated management system uses one set of processes for the shared elements, such as document control, internal audit, management review, competence, and corrective action, while keeping the content specific to each discipline. Integration saves effort, avoids duplicate and conflicting procedures, and gives leaders one view of the business. A change to a production line, for example, goes through one management of change process that considers safety, quality, food safety and environmental effects together.

Integration has risks. The discipline with the strongest team or the most external pressure can dominate. At many food companies, food safety and quality carry customer and regulatory weight that safety can lose out to. Safety has features the other systems lack, notably the requirement for worker consultation and participation, and an integrated system must not dilute them. Harrowgate Foods Ltd. kept the safety committees' role intact, kept a worker co-chair on the integrated management review, and kept safety-specific objectives separate from quality targets.

Capstone: a gap assessment

A gap assessment compares an organization's current system with the requirements of a standard or a target state, element by element, and identifies what needs to be done to close the gaps. It is usually the first step in building a system, and a useful health check for an existing one. Harrowgate Foods Ltd.'s next step was to acquire a small spice blending plant with about 40 workers, a site with basic safety compliance but no management system. Elena used the acquisition as the capstone of the company's learning: a gap assessment that pulled together every lesson.

The assessment rated each element on a scale of 0 to 4: 0 means absent, 1 means informal or partial, 2 means documented but not consistently implemented, 3 means implemented and effective, and 4 means implemented, effective and improving. Evidence came from documents, observation and interviews with a third of the workforce, across both shifts.

# Gap assessment for the acquired spice plant. Target maturity: 3.
scores = {
    "Context and scope": 1, "Leadership and policy": 1,
    "Worker participation": 2, "Hazard identification and risk": 1,
    "Legal and other requirements": 2, "Objectives and planning": 0,
    "Competence and training": 2, "Communication and documents": 1,
    "Operational control and change": 1, "Contractors and procurement": 1,
    "Emergency preparedness": 2, "Monitoring, audit and review": 0,
}
target = 3
gaps = sorted(scores.items(), key=lambda kv: kv[1])
print(f"Average maturity: {sum(scores.values()) / len(scores):.1f} of 4")
print("Largest gaps:")
for name, s in gaps[:4]:
    print(f"  {name}: {s} (gap {target - s})")

Average maturity: 1.2 of 4
Largest gaps:
  Objectives and planning: 0 (gap 3)
  Monitoring, audit and review: 0 (gap 3)
  Context and scope: 1 (gap 2)
  Leadership and policy: 1 (gap 2)

From gaps to a plan

A gap list is not a plan. Elena's team prioritized gaps by risk, not by score. The lowest scores were in objectives and in monitoring, audit and review, but these mattered less in the first months than the gaps that could hurt someone soon: the blending room's dust hazard had never been assessed, two mixers had no equipment-specific lockout procedures, and contractors who serviced the dust collector had never been prequalified. Those went first. The plan ran in three phases over eighteen months:

• First 90 days: assess and control the highest risks (dust, lockout, contractors), set up a joint occupational health and safety committee meeting schedule, introduce the hazard reporting system and the company's emergency plan template, and brief leaders on the policy.
• Months 4 to 12: build the risk register with workers, align procedures and training to company standards, add the site to the legal register, set objectives, and include the site in the internal audit program.
• Months 13 to 18: first internal audit, first management review input, and readiness for the site to join the company's external recognition scope.

The gap assessment also produced something less tangible. The workers at the spice plant had been wary of the acquisition. Being interviewed, seeing their concerns written into the plan and having two of their own trained as auditors in the first year did more to build trust than any announcement.