Why English? | Writing a Pharmaceutical Product Quality Review

A Product Quality Review, often shortened to PQR, is a periodic, structured look across a medicinal product’s manufacturing and quality evidence. It asks whether the process remains consistent, whether specifications and controls remain suitable, and whether recurring signals require improvement.

Singapore’s Health Sciences Authority states that it adopts the current PIC/S Guide to Good Manufacturing Practice for Medicinal Products as its GMP standard. The HSA GMP page is therefore the right starting point for current local context; a copied checklist from an unknown source is not.

• Confirm continued process consistency.
• Review trends rather than isolated numbers.
• Check whether starting-material and finished-product specifications remain appropriate.
• Connect quality events that may share a cause.
• Identify product, process or control improvements.
• Record commitments so they can be followed to completion.

Begin with product name, dosage form, strength, market, manufacturing sites, review period and responsible functions. State whether the review covers all batches manufactured, only released batches, validation batches, rejected batches, outsourced steps and products made for more than one market.

Scope language should reveal exclusions. “All relevant batches were reviewed” is circular because relevant is undefined. “All 42 commercial batches manufactured from 1 January to 31 December, including two rejected batches and three stability batches” is auditable.

A reliable PQR is assembled from controlled records: batch data, laboratory results, deviations, out-of-specification and out-of-trend investigations, complaints, recalls, returns, stability studies, change controls, supplier reviews, validation records and regulatory commitments.

List each source, system, query date, owner and completeness check. If systems use different product codes or time zones, reconcile them. A polished narrative cannot repair an incomplete population.

1. Define the data fields needed.
2. Identify authoritative systems and record owners.
3. Extract data using documented criteria.
4. Reconcile counts across sources.
5. Investigate missing or duplicated records.
6. Freeze the reviewed dataset or retain a traceable snapshot.
7. Record any late data and its impact on conclusions.

The writer may coordinate inputs, but quality, manufacturing, engineering, laboratory, supply and regulatory colleagues own different evidence. State who supplies, analyses, challenges, approves and tracks actions.

Avoid the passive voice when responsibility matters. “It was decided that monitoring would continue” hides the decision-maker. “The Quality Council approved monthly monitoring for six months; the Process Owner will report by 15 July” can be followed.

• PQR coordinator and document owner
• Data owners for each section
• Subject-matter reviewers
• Quality-unit approver
• Action owners and governance forum
• Escalation path for critical findings

Count batches manufactured, released, rejected, reprocessed, reworked, returned or still pending. Reconcile these totals before interpreting rates. A denominator must match the question: complaint rate may use distributed packs, while deviation frequency may use manufactured batches.

Describe meaningful production changes: scale, equipment, site, process parameter, campaign length or raw-material source. The PQR should help a reader understand whether this year is comparable with last year.

Process data may include yields, blend time, temperature, pressure, tablet weight, hardness, coating gain, fill volume or environmental conditions. Do not paste every chart into the narrative. Select the parameters that reveal control and explain why they matter.

Use a three-part sentence: claim, evidence, meaning. “Compression yield remained stable: all 39 released batches fell within the historical range, and no run rule was triggered; this supports continued process consistency for the period.” If the evidence is mixed, write a mixed conclusion.

• State the parameter and approved or statistical reference.
• Describe the direction, spread and unusual points.
• Link excursions to investigations or known changes.
• Distinguish common variation from a special cause.
• Avoid declaring capability without suitable analysis.
• State the monitoring or improvement decision.

Review critical quality attributes, assay, impurities, dissolution, microbiological results, physical tests and any product-specific measures. Include out-of-specification investigations and meaningful out-of-trend signals.

A retest that passes does not erase the original result. The report should state the investigation conclusion, scientific rationale, impact assessment and action. Avoid language that suggests testing continues until a preferred answer appears.

Stability studies show how product quality changes under defined storage conditions. Review protocol adherence, pull points, tested attributes, trends, excursions, failures and remaining shelf-life support.

Separate observed data from prediction. “Impurity A increased from 0.08% to 0.14% over twelve months” is observation. “The trend may approach the limit before expiry” is interpretation that needs an appropriate model, uncertainty and expert review.

• Batches and pack configurations on stability
• Storage conditions and time points
• Missing, delayed or repeated tests
• Statistical or graphical trend method
• Commitment and ongoing programmes
• Impact on shelf life, storage statement or specification

A finished product can drift because a raw material, component, supplier, test method or transport condition changed. Review critical materials, supplier performance, incoming-test results, complaints, audits and approved changes.

Do not blame a supplier before evidence supports the link. Write the sequence: signal, affected lots, investigation, supplier response, incoming controls and product impact. If causation remains uncertain, say so and identify the next test.

• Critical starting materials and packaging components
• New or alternate suppliers
• Specification or test changes
• Supplier deviations and corrective actions
• Rejected or conditionally accepted lots
• Trend across material lots and product outcomes

A deviation count is a starting point. Classify events by process step, cause category, product impact, recurrence, severity and closure status. Then look for clusters that separate records may hide.

Repeated “operator error” conclusions deserve careful challenge. Similar events may indicate confusing instructions, poor equipment design, workload, training design or weak detection controls. The language should not turn a system problem into a person label without evidence.

1. Reconcile all product-related deviations.
2. Group by meaningful mechanism, not convenient department.
3. Check repeat and related events.
4. Review overdue investigations.
5. Compare stated causes with CAPA effectiveness.
6. State residual uncertainty and next steps.

Quality events use specialised terms because they trigger different workflows. The PQR should preserve those distinctions while showing relationships. An OOT result may precede an OOS result; a process trend may explain both; a change may explain the process trend.

Build an evidence chain rather than a dramatic story. Time order alone does not prove cause. Compare affected and unaffected batches, alternative explanations and the effect of interventions.

HSA describes a Product Quality Surveillance Programme for therapeutic products. Post-market information belongs in the quality picture because patients and users encounter the product outside the factory. Consult the current HSA quality-surveillance page for official context.

Trend complaints by verified defect, presentation, batch, market, distribution volume and reporting delay. Separate medical adverse-event systems from product-quality complaint systems while linking them when procedures require.

• Complaint count and rate with denominator
• Confirmed, unconfirmed and inconclusive classifications
• Serious or critical events and escalation
• Returns and reasons
• Recalls, mock recalls and authority communications
• Recurring codes, batches, packs or markets

List significant changes to process, method, equipment, facility, software, supplier, specification, packaging or storage. For each change, state the rationale, implementation date, affected batches and effectiveness evidence.

“Change closed” is an administrative status, not proof of effectiveness. Ask whether the expected benefit appeared, whether new risks emerged and whether enough post-change data exists. If the evidence window is short, write an interim conclusion and a future review date.

Corrective and preventive action should address causes and reduce recurrence or risk. Review open, overdue and closed CAPAs linked to the product. Then test whether the effectiveness criteria were meaningful.

Training is an action, not automatically a complete solution. If a process design invites error, retraining alone may not control the risk. The report should show why the selected action matches the cause.

• Problem and verified or probable cause
• Action aligned to that cause
• Owner and target date
• Interim risk control
• Effectiveness measure and observation period
• Evidence of recurrence or sustained improvement
• Residual action if effectiveness is not demonstrated

A good conclusion answers the review questions. Was the process consistent? Are current specifications and controls appropriate? Are there trends or recurring problems? What improvements are required? Which uncertainties remain?

Avoid both false reassurance and exaggerated alarm. “All results passed, therefore the process is robust” may ignore a steady drift. “A trend was observed, therefore product quality is compromised” may overstate a weak signal. Calibrated language is a quality control.

Example: “The process remained capable of producing released batches within specification during the review period. A gradual increase in Impurity A across stability intervals warrants enhanced monitoring; no released batch exceeded the approved limit. The site will complete a formulation-and-packaging review by 30 September.”

An observation becomes useful when it has an owner, due date, priority, deliverable and completion evidence. Avoid “consider reviewing” unless consideration is genuinely the assigned task.

Prioritise by patient and product risk, not by how easy the action is. If an action spans sites, name a single accountable owner and supporting contributors.

A PQR may encounter unavailable records, late investigations, inconsistent codes or data extracted after the cut-off. Do not conceal these limits in an appendix. State what is missing, why, the potential effect and the recovery plan.

If two systems disagree, preserve both counts until reconciled. Choosing the convenient number creates false certainty. The discrepancy may itself expose a quality-system weakness.

• Nature and extent of the gap
• Products, batches or periods affected
• Interim risk assessment
• Compensating evidence reviewed
• Owner and due date for reconciliation
• Whether the conclusion is provisional
• Trigger for reopening or amending the review

A future reader should be able to reproduce the population, understand the analyses and see why actions were chosen. Use stable section names, controlled tables, traceable references and version history.

Keep raw listings accessible without burying conclusions. The main report should tell the quality story; appendices should support it. Cross-reference deviations, changes and CAPAs consistently so the evidence graph remains navigable.

1. Reconcile counts and denominators.
2. Check every chart label and unit.
3. Verify dates, batch numbers and record references.
4. Confirm conclusions match displayed evidence.
5. Ensure actions have accepted owners.
6. Resolve comments without deleting the audit trail.
7. Approve through the current quality system.

Professional quality writing uses verbs that reveal evidence strength: observed, indicated, confirmed, supported, suggested, excluded, reconciled and remained unresolved. Each verb should match the record.

Read a paragraph aloud. If the subject disappears behind passive constructions, restore it where accountability matters. Retain passive voice when the process or result, not the actor, is the useful focus.

1. Confirm current regulatory and site requirements.
2. Approve scope, product population and data sources.
3. Reconcile batches and core records.
4. Analyse manufacturing, laboratory, stability and supplier evidence.
5. Connect deviations, changes, complaints, recalls and CAPA.
6. Challenge conclusions against contrary evidence.
7. Assign risk-based actions with owners and dates.
8. Complete independent review and approval.
9. Track actions and carry open items into the next review.

The sequence is deliberately circular. The next PQR should not rediscover this year’s unresolved actions. It should show what changed, whether the action worked and what the longer trend now means.

Use the current HSA GMP standard page and PIC/S publications when checking requirements. Cycle-sensitive guidance should be dated in the report; do not rely on remembered wording.

Imagine a tablet product with twelve commercial batches in the review period. All were released. Three deviations concerned tablet-weight adjustment, two stability pulls were late, four complaints described chipped tablets, one supplier changed a coating-material manufacturing site and an assay trend rose slightly while remaining within specification. The exercise is to turn these facts into a review, not into a dramatic story.

Step 1: define the population and reconcile it

Write a scope sentence naming product, strength, dosage form, site and dates. Create a batch table with all twelve batches and show disposition. Reconcile twelve manufactured with twelve released. Record whether any validation, rework, rejection, return or pending batch exists. This may feel basic, but every later rate and trend depends on the population being stable.

Next, create a source register. For production, use approved batch and process records. For laboratory evidence, use the controlled laboratory system. For complaints, use the complaint database and distributed-unit denominator. For the supplier change, use the approved change-control record. State the extraction date and any known coding limitation.

Step 2: write three claims at different evidence strengths

First write an observation: assay results increased from the lower part of the historical range toward its centre across the twelve batches. Second write an interpretation: the direction warrants review because it coincided with a raw-material site change, although all results met specification. Third write a decision: continue enhanced assay and material-attribute monitoring for the next six batches while evaluating whether the site change plausibly affects the result.

Now challenge yourself. Coincidence is not causation. Were pre-change and post-change batches made on the same equipment? Did the method change? Is the apparent slope large relative to analytical variability? Does stability show the same direction? A good PQR records contrary evidence instead of arranging facts to confirm the first hypothesis.

Step 3: connect the chipped-tablet complaints

Do not write only “four complaints received”. Identify batch, pack, market, complaint verification and distribution denominator. Ask whether damage occurred in manufacture, packing, distribution or user handling. Compare friability, hardness, packaging-line events, shipping excursions and returned samples. If two complaints involve one batch, say so; if no sample was available, preserve that limitation.

The conclusion may be mixed: complaint frequency increased, returned-sample evidence is incomplete, batch tests met requirements and a packaging-line inspection identified one intermittent guide misalignment. The action can still be strong: repair and verify the guide, inspect retained samples, review transport evidence and trend the next production campaign.

Step 4: deal with late stability pulls honestly

Record planned and actual pull or test dates, the reason for delay, storage control and impact assessment. Avoid changing the label from late to “administrative timing difference” unless the procedure truly supports that classification. Check whether the two events share scheduling, capacity or system causes. Assign one action that addresses the mechanism rather than issuing a generic reminder.

Step 5: finish with a decision table

• Process state: what the twelve-batch evidence supports
• Laboratory state: assay direction, variability and monitoring decision
• Stability state: data support and late-pull limitation
• Complaint state: signal, evidence gap and packaging action
• Supplier change: effectiveness status and remaining observation period
• Open actions: owner, date, interim control and completion evidence

Read the finished conclusion aloud to a manufacturing colleague, a laboratory colleague and a quality colleague. Ask each what decision they think the paragraph requires. If they hear three different actions, the writing has not yet integrated the system.

• Write a laboratory method validation report for analytical evidence.
• Write a supplier nonconformance report for material and vendor issues.
• Write a product recall communication when market action is required.
• Write a root cause analysis report without overstating causation.
• Check HSA’s current GMP and GDP standards page.

A good PQR makes a large system legible. It respects the details, shows the pattern, calibrates the conclusion and leaves the organisation with a smaller, clearer set of actions that protect product quality and patients.