Y2K Revisited: A Human Component?

To the Editor: Few events in recent times have been as anticlimactic as the Y2K transition. Whether it was because of thorough preparedness or overstated worries about a largely nonexistent problem, January 1, 2000, came and went uneventfully in the electronic world. Yet, in all the attention about internal electronic dates, the possible effect of Y2K on the timing of human mortality may have been overlooked.

I determined the total number of deaths for the past 4 years that occurred each month at Yale-New Haven Hospital, exclusive of fetal deaths. Cause of death was determined from the death certificate and grouped into 1 of 10 categories. Statistical analysis of outlier months was performed using the Fisher protected least significant difference test.

The number of deaths per month at Yale-New Haven Hospital has remained relatively constant near a mean of 75. However, in January 2000, there were 123 deaths (Figure 1), which is more than 5 SDs above the mean monthly deaths for past 4 years (P<.001). Only 3 of the deaths occurred on January 1, and they were not related to equipment failure. The deaths were relatively evenly distributed over the month with 61 deaths on or before January 16th and 62 of the deaths afterward. There were no changes in hospital policies or staffing during this month. The age distribution of the deaths was essentially the same as other months, with a slightly higher representation in those aged 61 through 70 years. The causes of death (Table 1) were similar to those in January 1999 and January 2000 with an overrepresentation of deaths from chronic pulmonary disease and a slight underrepresentation from deaths due to acute vascular events (ie, myocardial infarctions, ruptured aneurysms, and strokes). Only 1 death certificate indicated influenza as a cause of death.

A less significant peak in the death count occurred in May 1997. That month had an unusually high number of medical examiner deaths (12%). Elimination of all medical examiner cases from the analysis decreased the variance in the number of deaths for May 1997 below statistical significance but increased the difference for January 2000.

A variation of 5 SDs in the number of deaths for January 2000 is unlikely to be random. The data also reveal a seasonal trend in the number of deaths, which parallels national statistics that show relative peaks in January and February and relative troughs in July and August.¹ During the peaks, death rates are typically 20% above the mean for the year. This trend is at least partly due to seasonal trends in influenza death rates and the January 2000 epidemic was particularly severe nationally.¹ The US mortality rate from influenza peaked at 11% in January 2000 compared with a mean mortality rate of 7% over the year.¹ However, a 4% increase in mortality due to influenza and a 20% increase in mortality from seasonal variation do not account completely for the 63% increase seen at Yale-New Haven Hospital during January 2000.

Although it is not possible precisely to explain the high mortality rate in January 2000, a likely contributing factor was the desire of patients to live into the next century. Most physicians have seen, at least anecdotally, the powerful effect of the patient's will to live, and a number of studies have supported a role for nonmedical factors in affecting patient outcomes.^{2 - 6} The overrepresentation of deaths due to chronic pulmonary disease and underrepresentation of deaths due to acute vascular events seen in January 2000 are consistent with this hypothesis. Being difficult to define, evaluate, or alter, "will to live" has been often disregarded as nonscientific. Yet, these data suggest a role for the patient's state of mind in postponing his or her own outcome.