When a Swimming Injury Log Comes Back Empty: The Most Dangerous Signal Is Missing Data

On June 12, 2026, at a 50-metre pool in Hai Phong, I received the weekly injury-tracking sheet for a group of 24 swimmers aged 15 to 19. The shoulder-status column read zero. The column for sessions missed due to pain read zero. The coach's notes column was left blank. The sheet was so clean I could read it in full sunlight without shading it with my hand.

I did not believe it. Not because I assume people lie, but because I knew where the denominator sat: the previous session was 6,000 metres of water, mostly freestyle, plus 40 minutes of resisted band work. At that load, a week without a single note is a phenomenon that needs explaining, not a result that deserves praise.

Two years later, an analysis file was placed in front of me. The title field: blank. The source field: blank. The list of information points: empty. All nine dimensions of deep professional analysis that the protocol requires simply could not run, and the final report contained exactly one honest sentence: insufficient information to assess.

The clean sheet at the Hai Phong pool and the empty file on the screen are the same class of event. Both are white data. In injury analysis, white data is the most dangerous kind, because it never incriminates itself.

Context: a sport with no medical memory

At Lach Tray, I learned to read injuries from the first numbers. In 2026, working as an assistant injury analyst for a football club, I built a simple training-load tracking system: every session, record volume, sprint minutes, and self-reported fatigue. Four months later the system flagged eight high-risk players before they entered a genuine injury phase. The lesson was not in the software. It was that data only has value when someone is willing to write it down every day, including the days when there is nothing to write.

When I moved to swimming, I realised that the thing I had taken for granted — a daily log sheet — is a luxury here. Vietnamese swimming has many pools, many training centres, many good coaches, and almost no centralised medical memory of athlete injury. No shared database. No shared definition of what counts as an injury case. No shared denominator to compare one centre against another, or one season against the next.

This is not uniquely Vietnamese. But in developed swimming nations, national federations have adopted injury-surveillance systems built on the International Olympic Committee standard published in the British Journal of Sports Medicine in 2026. That standard is explicit: an injury is any musculoskeletal complaint an athlete sustains in training or competition, regardless of whether it requires medical attention or causes time loss. A shoulder that aches but can still swim is, under that standard, an injury case. A shoulder that aches so much the athlete misses three sessions is also an injury case, at a higher severity.

In many places, only the second kind gets counted. The first kind is filed under "normal". And once the first kind is filed under normal, the clean sheet appears. Not because shoulders are healthy. Because nobody is listening.

Based on my experience tracking national meets and long training camps over nearly two decades, I would argue the biggest problem in Vietnamese swimming is not a shortage of machines or doctors. It is the absence of a habit of recording, and the absence of a person empowered to record without being treated as the bearer of bad news.

The shoulder in swimming is a structure under a particular kind of load. It does not absorb impact like a footballer's knee. It absorbs repeated rotation through a large arc, with the arm overhead, thousands of times per week. This injury does not arrive with a fall. It arrives as a graph. Every fall has a graph, and every graph has a breaking point — except in swimming, the breaking point usually sits in week nine of a training block, not minute ninety of a match.

Three mechanisms that produce a blank sheet

When an injury log comes back with zeros and an empty notes column, there are three possible explanations, and only one of them is good news.

The first mechanism is a collection failure. The recorder has never been trained on the definition, so they count by instinct. To a coach, "injury" usually means "cannot get in the water". To a doctor, it means "identifiable tissue damage". To a 16-year-old athlete, it means "pain severe enough that I can no longer hide it". Those three definitions sit far apart, and the gap between them is exactly the region of lost data.

The second mechanism is definition drift. Even when a single centre records for years, the threshold can shift when the person in charge changes. One recorder counts every mild case. Their successor counts only cases involving missed sessions. The injury curve suddenly drops, and the coaching staff celebrate a medical achievement that is really just a change in definitions. I have seen that kind of "improvement" at least twice in my career, and neither time was any athlete actually healthier.

The third mechanism is reporting pressure. This one is the hardest to discuss and the most expensive. When the person recording and the person being evaluated on results are the same person, or sit in the same structure, a recorded injury stops being a medical fact. It becomes a confession of coaching error.

These three mechanisms do not exclude each other. They compound. And after enough compounding, a centre ends up with a record that is clean, polished, and useless for any purpose whatsoever.

To be clear: I am not accusing any individual. I am describing system design. A system that asks people to self-report their own weaknesses, without separating professional evaluation from performance evaluation, will inevitably produce white data. Not because people are bad. Because people are rational.

The load arithmetic: more than a million shoulder rotations a season

To see why a blank sheet is absurd, you only need one multiplication.

Take a freestyle swimmer in the 15-to-19 age group training 6,000 metres per session, six sessions per week. Total volume is 36,000 metres weekly. In a 25-metre pool during base training, that is 1,440 lengths. A competent freestyle swimmer performs roughly 18 stroke cycles per 25-metre length, depending on height, reach and speed.

Multiply it out: 1,440 times 18 gives approximately 26,000 stroke cycles per week. A 48-week season yields around 1.24 million shoulder rotations. Each cycle involves one overhead reach, one internal rotation, and one pull against drag that scales with the square of velocity.

I do not need any further anatomy to see that a structure subjected to more than a million repetitions in a year requires regular measurement. Yet in most of the swimming centres I have worked in, that structure is monitored by a single question asked verbally after training: "How's the shoulder?"

That question is not data. It is a social ritual.

For comparison, Michael Phelps's peak-era training volume was widely reported in the international sports media at roughly 80,000 metres per week, based on descriptions by coach Bob Bowman. I raise that figure not to encourage anyone to copy it, but to make one point: at that level, volume is controlled by schedules, by testing, by logs — not by instinct. The difference between a leading swimming nation and a developing one is not metres swum. It is lines written down.

The body is a closed system, but data is the key that opens it. Without the key, all you can do is stand outside and guess from the noises within.

The forgotten denominator

There is one technical error I encounter again and again in injury reports sent up from units: they report the numerator and forget the denominator.

For example, a centre reports nine shoulder injury cases requiring three or more days off over 12 weeks. That sounds like a lot. But a lot compared to what? If the centre has 40 athletes training 20 hours in the water per week, total exposure is 800 water-hours per week, or 9,600 water-hours over 12 weeks. The rate is 0.94 cases per 1,000 water-hours.

Set that against a second centre with 40 athletes training only 12 water-hours per week: exposure of 5,760 hours, and if that centre records six cases, the rate is 1.04 per 1,000 water-hours. The second centre has fewer cases in absolute terms, but a higher injury rate.

Without a denominator, every comparison is a game of feeling. A low-volume centre will always look safer than a high-volume one, even when its prevention quality is markedly worse.

This is why I always demand two minimum data lines before debating anything: total exposure hours, and the injury definition in use. Missing either one, the debate is just an exchange of beliefs.

A four-stage pipeline and the break at stage one

In my analytical work I picture the injury data process as a pipeline of four sequential stages. Stage one is collection: recording raw events, every session, every athlete. Stage two is standardisation: bringing every record onto the same definition, the same unit, the same time frame. Stage three is cross-checking: against that athlete's own history, against the group baseline, against last season's baseline. Stage four is conclusion: issuing a risk judgement and converting it into a specific change in the training plan.

This pipeline only runs when all four stages have material. If stage one returns empty — no raw events recorded at all — stage two has nothing to standardise, stage three has nothing to compare, and stage four is forced into exactly one sentence: insufficient information to assess.

What matters is the response at stage four. There are two ways to handle an empty pipeline. The first is to stop, flag the error, and go back to repair stage one. The second is to fill the gap with conjecture and present that conjecture as a finding.

The second way is always faster. It is also always more wrong.

I once saw an internal report state that "the team's injury rate fell 40 percent year on year", when in reality the prior year had no complete recording at all. That 40 percent figure was not arithmetically wrong. It was simply calculated from a denominator that did not exist. This is the most dangerous class of error in sports analysis, because it dresses a gut guess in quantitative clothing.

The imputation trap

There is an occupational temptation every analyst has met: when the table is empty, you want to fill it in.

Fill it with an estimate from experience. Fill it with the group average. Fill it with last season's number. Each time you do, the table looks fuller, the report looks more professional, and the distance between the model and the actual body widens by another segment.

My rule is simple, and I have kept it for my whole career: no source, no number. An empty cell must be marked empty, not patched with a statistically plausible value. The difference between "no data" and "data equal to zero" is the difference between an open wound and a healed scar. They look alike at a glance. Handle them wrongly and the consequences diverge sharply.

This is especially true in youth swimming. A 15-year-old swimmer in a growth phase can change height by several centimetres in a single season. Muscle, tendon, ligament and cartilage develop out of step with each other. A model built on last season's figures applied to this season's body is a wrong model — not because the mathematics is wrong, but because the subject has changed.

That is why I accept a "good enough" standard in data collection. Waiting for a perfect dataset before writing the first line is the surest way never to write a line. Incomplete beats fabricated. Wrong-day beats wrong-month. I set a deadline for collection, and I accept that my dataset will always have holes.

The counter-angle: a clean sheet is scarier than a messy one

Here I go against the crowd's default reflex. When a centre announces "no injuries all season", the default reaction is praise. My reaction is to question the thermometer.

A log with a few messy notes, a few shoulder aches of unclear origin, a few athletes who missed two sessions and came back — that log looks bad. But it is honest. It tells me the system is working, that somebody is listening to athletes' bodies patiently enough to record small things.

A perfectly clean sheet has two explanations. The first: the centre has reached a level of prevention where injuries are intercepted before they arise. The second: the thermometer is broken. Statistically, the second is far more common, especially where no standardised surveillance exists.

The problem is that you cannot distinguish these two explanations by looking at results. To distinguish them you must audit the process: is anyone recording, under what definition, is there any cross-checking, is the recorder penalised for the number of injury cases.

In other words, the right question is not "how many injuries". The right question is "who recorded, and how".

I understand why this view is unpopular. It turns a good media story into a paperwork audit. But if I chose the crowd's framing, I would have to write that a sport involving more than a million shoulder rotations per athlete per season produced not a single shoulder complaint worth recording. I cannot write that sentence.

Closing

At Lach Tray, I learned to read injuries from the first numbers. Numbers are silent, but their sequence always knows how to tell a story. And the sequence in Vietnamese swimming is telling a story that has not yet been written down: thousands of young athletes, tens of thousands of water-hours each week, and almost no data patient enough to stay across seasons.

Hai Phong, Moscow and COVID — three milestones that taught me injuries are never identical. Every body breaks in its own way. But the way we fail to see it is always the same: an empty cell left alone, a question asked carelessly, a clean sheet rewarded.

The question I leave with those working in youth development is not how to reduce injuries. The question is this: if next season your tracking sheet comes back with zeros again, will you treat it as good news — or will you go and check whether the pen still has ink?