How to Pass the CPHQ Exam on Your First Attempt

Almost every unsuccessful first attempt I see is a method failure, not a knowledge failure. The candidate knew what a control chart was, could recite the steps of a root cause analysis and had read the textbook twice. What they had never practised was choosing between four defensible actions in a scenario where three of them are things a competent quality professional might reasonably do, and only one is the right thing to do next.

NAHQ does not publish a pass rate, and any site quoting one should be treated with suspicion. What NAHQ does publish is more useful: the split of item types. Twenty-three per cent recall, 57% application, 20% analysis. Read that as an instruction rather than as trivia. Three-quarters of the exam asks you to do something with knowledge instead of stating it.

The people who find this hardest are strong clinicians with little improvement-project experience. They have the vocabulary and not the reflexes: they can define a driver diagram but have never had to decide whether to build one or to go and watch the process first.

You receive a scaled score between 200 and 800, and 600 passes. A scaled score is not a percentage and cannot be converted into one. Fifteen of the 140 items you answer are unscored pretest questions, and you will not be told which fifteen. So stop asking how many you are allowed to get wrong.

NAHQ reports the scaled score and the passing standard, not a raw-item equivalent. — So a mock percentage is a training signal, not a prediction. I treat a steady 75% to 80% across mixed-domain mocks, with errors spread evenly rather than piled into one domain, as the signal to book the date.

Domain names tell you nothing about how much of your score they control. The item counts do, and NAHQ publishes them. The seven domains are set out in full in the CPHQ exam blueprint; the table below is the version you should study from.

Improvement and analytics together are 53 of the 125 scored items, roughly 42% of your score. Regulatory and accreditation is the domain candidates most often over-prepare, because standards are concrete and revising them feels productive. Eight items do not justify a fortnight.

NAHQ suggests between 20 and 50 hours of preparation, and most candidates already working in healthcare quality report six weeks to three months. Forty hours suits someone with a few years of quality experience; if you are new to improvement work, plan for the upper end, as the eight-week CPHQ study plan does.

An application stem has four parts: a setting, a trigger, a constraint and the actual question. Candidates lose marks by answering the trigger. Read the final sentence first and note its qualifier, because first, next, best and most likely point at different options within the same set of four.

A medical ward reports five wrong-time medication administrations in one month. The unit manager asks you to deliver a training session for the nurses involved. You have the five incident reports and no other data. What is your first action?

The answer is B. With five events and no process data, characterise the failure before choosing a countermeasure. A fails because it commits to the weakest intervention available, education, before anyone knows what the problem is; the manager's request is the trap. C fails because measurement without a change is surveillance. D fails because it alters the process on an untested assumption about cause.

A run chart of monthly catheter-associated urinary tract infection rates shows nine consecutive points below the median, starting the month after a new insertion checklist was introduced. The team asks whether the improvement is real. Which interpretation is most appropriate?

The answer is B. Run-chart rules treat six or more consecutive points on one side of the median as a shift, a signal of non-random variation, and nine points clear that threshold. A fails because it dismisses a signal the rules already identify. C fails because recalculating the median across the changed period removes the evidence you are testing; you re-baseline after improvement is sustained, not to decide whether it happened. D fails on sequence rather than truth: a control chart adds precision, but the run rules already answer the question.

You lead a hospital-wide sepsis bundle project. Compliance is above 90% on two wards and 40% on a third, whose clinical lead has missed the last four project meetings and reports to a different director. You have no line authority over that ward. What is the best next step?

The answer is B. Influence without authority works through the other party's own forum and their own version of the problem, and ward-level data shown where their peers sit supplies both the reason and the route. A fails because escalating first spends credit you will need later and turns a colleague into a defendant. C fails because reporting only the compliant wards abandons the patients most at risk. D fails because it assumes the barrier is ignorance, when four missed meetings suggest priority.

On a recall item you either know it or you do not. On a scenario item you can reason your way there even when two options look equally sound. This is the sequence I teach, and it should take under a minute.

A question bank is a diagnostic instrument that most candidates use as a scoreboard. The wrong answers tell you everything the score does not. Work in sets of 20 to 30, mark them immediately, and log every error under one of four causes.

Never re-sit a set twice, or you are testing memory of the item rather than the reasoning. For fresh scenarios with the reasoning attached, work through the fifteen fully explained CPHQ questions and then sit the free CPHQ practice test under timed conditions.

In candidates who fail once and then pass comfortably, nothing usually changed about their knowledge between the two attempts. The first-attempt preparation has one shape: read the textbook end to end, build a large set of flashcards, save the practice questions for the final week, and arrive knowing a great deal of content they had never been asked to use. The retake is shorter and mostly items and error analysis, with reading used only to repair specific gaps.

Three hours for 140 questions is about 77 seconds an item, comfortable for recall items and tight for analysis items with a data table. The pace to hold is a quarter of the paper every 45 minutes, which leaves roughly 15 minutes at the end for the items you flagged.

Two rules go with it. Never leave an item blank, and always commit to an answer before you flag it, so that running out of time costs you nothing.

Sit your last full-length mock five to seven days out, not the day before: you need time to act on the two weakest domains it exposes. The final three days are for your own error log and your own domain summaries, and for nothing new.

Confirm the practical arrangements early rather than on the day: whether you are sitting in person or online with a live remote proctor, what identification you must present, and, for the online option, the system check. — On the morning, eat properly, arrive or log in early, and plan to use most of the three hours. Candidates who finish in 90 minutes usually did not read the qualifiers.

Self-study is sufficient if three things are true: you have run improvement projects yourself, you can read a control chart without hesitating, and you can hold a schedule unsupervised. Buy a textbook and a large question bank, follow the weights above, and you will be fine.

A course earns its place when one of those is missing, and it is usually the second. Analytics is where clinicians lose marks and the hardest domain to teach yourself from a book, because the skill is interpretation rather than recall. If you want that structure, our Complete CPHQ Prep course is a single payment of $399 with no subscription and no expiry, covering 140+ video lessons, 1,000+ practice questions and full-length mock exams with lifetime access.

Either way, the test of readiness is the same. Sit a mixed-domain mock under timed conditions and read your error log rather than your score. If the errors are spread across domains, are mostly close pairs rather than blanks, and you can say in one sentence why the correct option beat yours, book the date.