For couples trying insemination, the city's carbon monoxide level may be a factor worth asking about

A fertility clinic in Zhongshan kept two kinds of records

A fertility clinic in Zhongshan kept two kinds of records, and the value came from putting them side by side. The first set tracked 2,328 intrauterine insemination cycles at the Reproductive Centre of Zhongshan City People’s Hospital. Those cycles ran from January 2014 through December 2025, so they cover a long stretch of ordinary clinic work. The second set came from air pollution readings in the central urban area during the same months. Matching them meant each insemination cycle could be linked to the air that patients breathed nearby. That design helps because a single clinic record cannot say whether outdoor gases were part of the picture. By comparing cycle outcomes with local readings, the analysis could test whether pollution levels moved in step with pregnancy results. The comparison then used pattern-finding computer models, together with gene pathway checks, to look for links. Those tools are not a lab experiment; they search large tables of existing records for signals that would be hard to see by eye. Because the clinic data were reviewed after the fact, the records show association rather than proof that a gas caused a result. The hospital setting also gives a clear place for the air data, because both records cover the same city and the same months. Taken together, the two record streams turn a local fertility clinic into a natural test site for asking whether outdoor air belongs in the story.

One pollutant stood out when the cycles were matched with air readings

When 2,328 insemination cycles from Zhongshan were matched with air readings from the city’s central urban area between January 2014 and December 2025, carbon monoxide stood out. It was the pollutant whose level during a specific follicle window carried useful information about clinical pregnancy. The window ran from the antral follicle stage to the mature follicle stage, when eggs are preparing for insemination. In a screening step that keeps only signals with real predictive weight, carbon monoxide retained a non-zero coefficient. Other air pollutants were pushed toward zero by the same calculation. In a separate prediction model, carbon monoxide ranked among the two most important predictors of clinical pregnancy during that window. Adding air-pollution variables made the models sharper than versions without them. They identified more true pregnancies and missed fewer, improving precision, recall, F1 score, and area under the precision-recall curve. This improvement means the air readings added information that cycle records alone did not provide. The result is an association, not proof of cause, because it used existing clinic and air readings. Gene pathway analyses pointed to inflammation and immune-related pathways as possible biological routes behind the link. Those pathways suggest a way in which exposure could be connected with lower clinical pregnancy rates.

As carbon monoxide rose, the chance of clinical pregnancy fell

During the window when a cycle moves from the antral follicle stage to the mature follicle stage, carbon monoxide in Zhongshan air tracked with lower pregnancy chances after insemination. A retrospective machine-learning analysis of 2,328 intrauterine insemination cycles at Zhongshan City People’s Hospital linked each cycle to local pollution readings from January 2014 through December 2025. The clearest pattern came from a flexible curve fit that allowed the relationship to bend rather than force it into a straight line. As carbon monoxide concentrations rose, the modeled clinical pregnancy rate fell from 16.09 percent to 8.50 percent. That is not a claim that one breath of gas caused a missed pregnancy. It is an association seen across many cycles and many days of air quality data. The drop was large enough to matter clinically, because it moved the expected success from about one in six cycles to just over one in twelve. The same inverse link appeared when the analysis used equations that adjust for repeated cycles within patients. A shrinkage step also kept carbon monoxide with a non-zero effect instead of discarding it as noise. In other words, the data did not point to a flat line where pollution made no difference. They pointed to a downward slope during one specific treatment phase. The timing is important because follicle growth creates a short biological window when the body may be more sensitive to external insults. The curve did not say that every extra unit of carbon monoxide cuts pregnancy by a fixed amount. It said the chance was highest at lower concentrations and lowest at higher ones, with the decline concentrated as levels climbed. That pattern is what makes the result harder to dismiss than a single average difference.

A small window with a large question

The finding sits at the edge of assisted reproduction, where a quiet gas becomes part of the story. A retrospective match of 2,328 insemination cycles with city air readings from Zhongshan between 2014 and 2025 found that carbon monoxide levels during one short stretch of treatment moved against clinical pregnancy as they rose. The expected clinical pregnancy rate fell from 16.09 percent to 8.50 percent, a drop that moved the chance from about one in six cycles to slightly more than one in twelve. That association is narrow enough to be trusted and broad enough to worry. It points to a time when the body may be especially open to what it breathes, and one stage can shape whether an embryo takes hold.

The wider lesson is that fertility care has long treated the clinic as a sealed room. The result asks for a different map: the air outside the door belongs in the chart, and the calendar of treatment may matter more than the average year. If carbon monoxide truly shapes this window, clinics could track local readings during the follicle transition, warn couples when levels climb, or adjust timing where possible. That would be a modest change with a large emotional weight, because many couples already feel that their outcome depends on things they cannot control.

The obstacles are real, and they decide how far this clue can travel. The finding shows an association, not proof of cause, and it comes from one hospital in one city over twelve years. Gene pathway analyses point toward inflammation and immune responses, but those pathways remain clues rather than settled biology. Future tests will need other places, cleaner air comparisons, and ways to separate carbon monoxide from the traffic, heating and industry that travel with it. The question left behind is simple: if a few days of treatment can be shaped by the air outside, who should decide when to move them?