Surgery chapters
Surgical audit, research, ethics, law, consent, human factors and patient safety
The audit cycle and how audit differs from research, basic study design and statistics, valid consent and capacity, confidentiality, and how human factors, checklists and quality improvement keep surgical patients safe.
- Audit measures care against an agreed standard and closes the loop by re-auditing after change; research creates new, generalisable knowledge and needs ethics committee approval.
- Valid consent needs capacity, voluntariness and enough information, including any material risk a reasonable patient would want to know; a signed form is not proof of consent.
- Most adverse events come from systems, not lone individuals (Swiss cheese model); the WHO surgical safety checklist (sign in, time out, sign out) reduces deaths and complications.
On this page
Overview
Good surgery is not only good technique. Surgeons must measure their results (audit), add to knowledge through research, respect patients' autonomy and confidentiality, and work in safe systems that catch human error before it harms. The UHS TOS asks for one MCQ on ethics in every paper.
Audit, service evaluation and research
| Feature | Clinical audit | Service evaluation | Research |
|---|---|---|---|
| Question | Does this service reach a set standard? | What standard does this service achieve? | What is new, generalisable knowledge? |
| Standard | Measures against a standard | No standard used | Tests a hypothesis |
| Treatment allocation | None; usual care only | None | May allocate or randomise patients |
| Data | Mostly routine | Mostly routine | Often extra tests or data |
| Ethics committee | Not usually needed | Not usually needed | Required |
Audit compares the structure, process and outcome of care with explicit criteria. If care falls short, change is made at the right level: an individual (training), team, hospital (e.g. new antibiotic policy) or nation (screening programmes). Topics often arise from morbidity and mortality meetings.
The audit cycle
- Choose the audit question (ideally with the multidisciplinary team).
- Find the evidence and set the standard.
- Design the audit and get institutional agreement.
- Collect data over an agreed period.
- Compare performance with the standard.
- Gap analysis: if standards are met, re-audit later; if not, agree changes (e.g. training).
- Re-audit to close the loop.
Audit is a circle, research is an arrow: audit keeps coming back to the same standard; research points to new knowledge.
An audit that is never repeated after change has not "closed the loop" and has not improved care.
Research basics
Types of study
| Type | What it does |
|---|---|
| Case–control | Patients with a disease compared with matched controls; looks back at exposures |
| Cohort (longitudinal) | Group followed over time (prospective or retrospective) |
| Cross-sectional | Measurements at a single point in time |
| Randomised controlled study | Patients allocated at random to new treatment or standard (control); best design for testing treatments |
| Systematic review and meta-analysis | Pools all good studies statistically; top of the evidence hierarchy |
Patient-reported outcome measures (PROMs) are now a key outcome in surgery.
Bias and blinding
- Blinded observer: the person recording outcomes does not know the treatment.
- Single blind: the patient does not know.
- Double blind: neither patient nor researcher knows until the study ends.
- Randomisation removes selection bias. In surgery it is not always possible or ethical.
- Publishing only positive results causes publication bias; negative studies should also be reported.
Errors, power and sample size
| Error | Meaning | Usual cause |
|---|---|---|
| Type I (alpha) | A difference is found that is not real (false positive) | Chance, many end-points tested |
| Type II (beta) | A real difference is missed (false negative) | Too few patients (underpowered) |
Wrong sample size is probably the commonest reason a study is invalid. Studies are usually designed with 80% power and a 5% significance level. Recruit more than the calculated number to allow for deaths, drop-outs and loss to follow-up; multicentre studies help.
Formula: n per group = 8 × [(100 − r)r + (100 − s)s] ÷ (r − s)²
Wound infection expected to fall from r = 30% to s = 10%
8 × [70 × 30 + 90 × 10] ÷ 20² = 8 × 3000 ÷ 400 = 60 patients in each group
Simple statistics
- Mean = average; median = middle value; mode = commonest value. In a normal (bell-shaped) distribution all three are equal.
- P < 0.05 is usually taken as significant: there is a 1 in 20 chance that the difference arose by chance alone.
- Confidence interval is the best guide to the likely true difference; a 95% CI for a difference that includes zero means no significant difference.
| Data type | Example | Test (two groups) |
|---|---|---|
| Numerical, normal | Blood pressure | t-test (unpaired; paired for before–after) |
| Numerical, not normal | Tumour size | Mann–Whitney U (paired: Wilcoxon signed rank) |
| Categorical | Admitted to ICU or not | Chi-squared test |
Research on patients needs approval from an independent research ethics committee and informed consent; the potential benefit must clearly outweigh known risks.
Ethics and law
Core principles
- Autonomy: a capable adult decides what is done to their body.
- Beneficence: act in the patient's best interests.
- Non-maleficence: avoid harm.
- Justice: fair, unbiased use of care and resources.
Consent
For consent to be valid, the adult must:
- have capacity for that decision;
- decide voluntarily (not coerced by family or staff);
- be given enough information.
Information to give: the diagnosis and its uncertainties; what the operation involves and why; risks and complications; likely outcome; alternatives, including no treatment; chance of success of each option; the responsible surgeon; and that they may change their mind at any time.
Material risk: following Montgomery v Lanarkshire, the surgeon must disclose any risk that a reasonable patient in that situation would regard as significant, or that this particular patient would be likely to care about.
| Legal claim | When it arises |
|---|---|
| Battery (assault) | Touching or operating without any consent, or doing a procedure not agreed to |
| Negligence | Consent obtained, but risks not adequately explained, or care fell below an acceptable standard and caused harm |
A signed consent form is only evidence that consent was sought; it does not prove the patient understood. Write a note of the risks discussed.
If a patient consented only to one procedure, do not add an unrelated operation "in their best interests" just because they are asleep, unless it is needed to save life or prevent serious harm.
Capacity
Capacity is decision-specific. A person has capacity if they can understand the information, retain it long enough, weigh it up, and communicate a decision. An unwise decision does not by itself prove a lack of capacity. A capable adult may refuse any treatment, even life-saving, e.g. a Jehovah's Witness refusing blood.
Special situations
| Situation | Approach |
|---|---|
| Unconscious emergency (no relatives) | Operate under the doctrine of necessity in the patient's best interests to save life or limb; honour a valid, applicable advance decision refusing that treatment |
| Temporary incapacity (e.g. intoxication) and surgery not urgent | Postpone until capacity returns |
| Children | Parent or person with parental responsibility consents; involve the child. A Gillick-competent child can consent (under English law cannot always refuse until 18) |
| Adults lacking capacity long term | Act in best interests, considering their past wishes and those close to them |
| Family asks you to hide a diagnosis | A capable patient who asks has a right to know; speak to the patient |
| Withdrawing life-sustaining treatment | Lawful when it is no longer in the patient's best interests; deliberately ending life (euthanasia) is unlawful |
Confidentiality
Patient information may be shared within the treating team when relevant (implied consent). Otherwise it is private, but not absolute:
- Must be breached: court orders, public health law (notifiable diseases).
- May be breached: to prevent serious crime or protect identified people at risk of serious harm.
Pakistani law and PMDC guidance broadly follow these principles, but details (age of consent, notifiable diseases, reporting duties) differ; check current national guidance.
Human factors and patient safety
Human factors is the study of how people interact with their tasks, equipment, environment and each other. The WHO estimates that about 1 in 10 patients is harmed by health care.
SHELL model
The human (Liveware) sits at the centre and interacts with Software (protocols, checklists), Hardware (equipment design), Environment (noise, crowding) and other Liveware (team communication).
Crew resource management training, borrowed from aviation, teaches situation awareness, decision making, communication, leadership and the effects of fatigue.
Definitions
| Term | Meaning |
|---|---|
| Adverse event | Incident that harms the patient |
| Near miss | Could have caused harm but was caught before reaching the patient |
| No-harm event | Reached the patient but caused no injury |
| Never event | Serious, wholly preventable incident, e.g. wrong-site surgery, wrong patient, retained swab or instrument |
Models of error
- Heinrich's pyramid: for every 300 near misses there are about 29 minor injuries and 1 major injury. Near misses are the best data on how safe a system is, so report them.
- Reason's Swiss cheese model: each defence is a slice with holes. Active failures (slips, lapses, mistakes by front-line staff) and latent conditions (management decisions such as understaffing, time pressure) create holes; harm occurs when the holes line up.
- System approach: ask why the system failed, not who failed. Most adverse events are due to system factors.
Errors can be slips/lapses (right plan, poor execution or memory) or mistakes (wrong plan). Fatigue, poor handover and hierarchy make all of them more likely.
The WHO surgical safety checklist
| Phase | When | Key checks |
|---|---|---|
| Sign in | Before induction of anaesthesia | Patient confirms identity, site, procedure, consent; site marked; anaesthesia check done; pulse oximeter working; allergy; difficult airway or aspiration risk; expected blood loss over 500 mL (over 7 mL/kg in a child) |
| Time out | Before skin incision | Team introduce themselves by name and role; confirm patient, site, procedure; critical steps, duration, expected blood loss; antibiotic prophylaxis given within 60 minutes; essential imaging displayed |
| Sign out | Before the patient leaves theatre | Procedure recorded; instrument, swab and needle counts correct; specimen labelled; equipment problems; key recovery concerns |
In the WHO study across eight hospitals, the checklist deaths fell from 1.5% to 0.7%; major complications fell from 11% to 7%.
A checklist only works if the team's attitude changes too; ticking boxes without real checks gives a false sense of safety.
Being open
After an adverse event, patients should get an honest explanation and apology (open disclosure, the duty of candour). Reports should be made without fear of blame so the system can learn.
Quality improvement
QI is continuous, data-driven change to improve care (audit is one tool within it).
- PDSA cycle (Plan, Do, Study, Act): test small changes quickly and repeatedly.
- Tools: root cause analysis, process mapping, fishbone (cause-and-effect) diagrams, driver diagrams.
- Measures look at structure, process and outcome; leadership and system thinking are essential.
Expect: audit versus research (standard, re-audit, ethics approval), the steps of the audit cycle, randomised studies and blinding, type I and II errors and sample size, P-values and confidence intervals, choice of simple statistical test, elements of valid consent and material risk, capacity, emergency treatment without consent, Gillick competence, Jehovah's Witness refusal, exceptions to confidentiality, Swiss cheese model, near miss versus adverse event, never events, and the three phases of the WHO checklist.
Quick revision
- Audit = measure against a standard, change, re-audit; no ethics committee needed usually.
- Research = new knowledge; needs ethics approval and consent.
- Type I = false positive; type II = false negative (small sample); power 80%, alpha 5%.
- CI for a difference crossing zero = not significant.
- Valid consent: capacity + voluntary + informed (material risks); form is not proof.
- No consent at all = battery; poor risk disclosure = negligence.
- Unconscious emergency: necessity and best interests.
- Swiss cheese: active failures + latent conditions; most errors are system errors.
- WHO checklist: sign in (before induction), time out (before incision), sign out (before leaving).
Written by The House Job from standard references (Bailey & Love).
FAQ
What is the main difference between audit and research?
Audit checks whether current care meets an agreed standard and is repeated after change; research seeks new, generalisable knowledge and needs ethics committee approval.
Is a signed consent form proof that consent was valid?
No. It is only evidence that consent was sought; validity depends on capacity, voluntariness and adequate information, so the discussion should be documented.
Can a surgeon operate on an unconscious patient without consent?
Yes, in an emergency, to save life or prevent serious harm, acting in the patient's best interests, unless a valid advance decision refuses that treatment.
What happens at the "time out" of the WHO checklist?
Before the skin incision, the team introduce themselves and confirm the patient, site and procedure, review critical steps and blood loss, and check that antibiotic prophylaxis has been given within 60 minutes.