METHODS

We took the results of the 1094 first team competitive football matches played by Newcastle United FC, Sunderland AFC, Middlesbrough FC, and Leeds United FC between 18 August 1994 and 28 December 1999 from Rothmans Football Yearbook.^9–14 For each team, match results were grouped as either a home defeat or any other combination of result and venue.

We examined the effect, using t tests¹⁵ of home defeat on the mean number of deaths on match days from acute myocardial infarction (ICD9 410/ICD10 I21–22) or stroke (ICD9 430–438/ICD10 I60–69) in the health authority area in which these clubs are based: Newcastle and North Tyneside, Sunderland, Tees, and Leeds Health Authorities respectively. Mortality data came from the Office for National Statistics. We compared the effect of home games (all results) against away games (all results), and we compared the effects of won, drawn and lost games, regardless of venue, on the mean number of male deaths from acute myocardial infarction and stroke.

We used Poisson regression¹⁶ to estimate for each health authority population and for all four combined the relative risk of dying from acute myocardial infarction or stroke on days when the local football team lost at home compared with days when any other result was achieved. Relative risks with 95% confidence intervals (CIs) were estimated separately for men and women, and were adjusted for the covariates month of year and day of week.

We compared the mean number of deaths from acute myocardial infarction or stroke on days when teams lost at home with the mean number of male deaths from these causes on the preceding and following three days.

RESULTS

Table 1 shows the mean number of male and female deaths per match day from acute myocardial infarction or stroke when the local professional football team lost at home, compared with days when the team played in a match with any other combination of result and venue. In each of the health authority areas, the average numbers of male deaths from acute myocardial infarction and stroke were greater on days when the local football team lost at home than on days when another result was achieved. In Sunderland Health Authority, stroke and acute myocardial infarction deaths increased by 66% in men on days when Sunderland AFC lost at home and this difference was significant.

Across the four areas combined (total resident population just over two million), male deaths increased by 30% when the local professional football team lost at home and this difference is highly unlikely to be attributable to chance alone. In contrast, the average numbers of female deaths were about the same on days when the local professional football team lost at home as on days when any other result was achieved.

The separate effects of home games and lost games on the rate of male mortality from acute myocardial infarction and stroke were much less powerful than the interactive effect of these two factors in combination. Across all four areas the mean daily number of male deaths from acute myocardial infarction and stroke increased very slightly, but not significantly, on days when the local professional football team played at home compared with when they played away (1.79 v 1.69, p=0.23). Across all four areas the mean daily number of male deaths from acute myocardial infarction and stroke increased slightly on days when the local professional football team lost compared with when they won (1.84 v 1.67 p=0.10).

Regression coefficients were adjusted for the covariates month of year and day of week to remove potential bias from this source. Table 2 shows the relative risks of death (and 95% CIs) from acute myocardial infarction or stroke on days when the local professional football team lost at home compared with days when other results were achieved.

The regression results agree with the comparison of means. For men living in each of the four areas, the risk of death from acute myocardial infarction or stroke was greater on days when the local professional football team lost at home. On days when Sunderland AFC lost at home, male residents of Sunderland Health Authority were 1.63 times (95% CI 1.16 to 2.3) more likely to die from these causes.

Across the four areas combined, male deaths from acute myocardial infarction or stroke increased by 28% (95% CI 11% to 47%) when the local professional football team lost at home.

Figure 1 shows the mean number of male deaths from acute myocardial infarction and stroke across the four areas combined on days when the local professional football team lost at home and for the preceding and following three days. This is compared with the mean daily number of male deaths from acute myocardial infarction and stroke in the four health authority areas for the period 1994–99, which was 1.71 (95% confidence interval 1.68 to 1.74).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Mean number of male deaths on days before and after the day of matches lost at home by the local professional football team, pooled across four health authority areas. *Horizontal line represents mean daily number of deaths in the four health authority areas combined for the period 1994–99.

On average, the number of male deaths from acute myocardial infarction and stroke was greater on days when the local professional football team lost at home than on the preceding and following three days. On average across the four areas combined, there were 0.54 fewer deaths on the day following football matches lost at home than on the day of football matches that were lost at home.

DISCUSSION

We have found an association between the outcome of football matches and deaths from circulatory disease, for men but not women. The association relates to all competitive games played by the local professional football team over a five year period. It extends the observation of Witte et al⁵ that male mortality increased after a single important international match involving Holland, and allows for possible distortion by the effects of differential death rates by day of the week and month of the year.

We selected the four health authority areas because of the characteristics of the local professional football club. They are the largest clubs in this health region, and the only ones that spent the whole study period within the top two levels of professional football. In each case, the great majority of the football support within the population covered by the local health authority goes to that club. The effect was seen consistently in each area studied and, although the observed increase achieved conventional statistical significance in only one area, the results in all areas combined were highly unlikely to be attributable to chance alone.

We took a home defeat as the critical outcome for local supporters, to be compared with other outcomes, for two reasons. Firstly, this is widely accepted to be the worst result for a team; supporters have lower expectations of their team’s performance away from home, but are predisposed to anticipate a victory in a home game. Secondly, more people will have attended a home game and witnessed the defeat at first hand.

Cardiovascular death rates are known to vary according to day of the week and month of the year. However, we removed the potential confounding effects of day of the week and month of the year, and we are unaware of remaining sources of bias in the analysis. On the other hand, our results probably underestimate the effect of a home defeat on supporters. Not all of the local health authority’s male population will follow football, possibly diluting the observed effect. Some may follow other clubs, perhaps even outside the area, potentially further reducing the observed effect. The true relative risk of death from circulatory disease immediately after a home defeat among known supporters of that team is likely to be higher than the observed value for the whole population. Because of the limitations of available data, our results are unable to differentiate between match day deaths occurring before and after the kick off. Given that it is highly implausible to suppose that there is some other factor causing a raised death rate in association with a home defeat that operates before a match has taken place, this is likely to represent some further underestimate of the strength of association.

The reasons why one person with underlying coronary artery or cerebrovascular disease may suffer an acute episode while another does not are not clear. The daily, weekly, and seasonal variations in the onset of acute coronary syndromes and sudden cardiac death suggest that these are “triggered events”.¹⁷ Trigger factors including unaccustomed physical exertion, emotional stress and external agents may act through haemodynamic effects, arrhythmias, hypertension, and hyperventilation,⁸ causing arterial plaques to become unstable and then disrupted. Emotional and psychological stresses that have been linked in this way with acute coronary syndromes and myocardial ischaemia include anger, sadness, frustration, and tension.⁶ Stroke rates are also known to be higher in association with mental stressors, particularly depression.⁷ Trigger factors have been estimated to play a causative part in 20% of cases of acute coronary syndrome.¹⁸

Our results do not identify the trigger factors that may have provoked the additional deaths observed. Losing at home may lead to anger, frustration, or depression among supporters, which are known to increase the risk of acute coronary and cerebrovascular events including sudden death. The Determinants of Myocardial Infarction Onset Study estimated the relative risk of acute myocardial infarction in the two hours after an outburst of anger to be 2.3 (reduced by aspirin to 1.4).¹⁹ Further study would be necessary to try to disentangle the effects of these potential trigger factors, and to shed light on why the association was found in men but not women.

It seems likely that the excess mortality that we have observed in association with a home defeat represents deaths brought forward that may otherwise have occurred later. However, there is no sign in our figures of a deficit in the days immediately after the peak, suggesting that this may be a comparatively long term effect. Our results therefore point to an area of potential public health improvement, as well as an opportunity to understand better the factors that may precipitate acute coronary and cerebrovascular disease.

Those with personal experience of the passion generated among fans devoted to a particular team may not be surprised by these results, having seen at first hand the effects of defeat. Others may question why disappointed fans do not simply switch to a more successful side. Although some do, to the true supporter this would be unthinkable. Pending further study, and given the evidence of its efficacy and relative safety, an appropriate preventive measure might be to recommend half an aspirin tablet to male supporters whose team has just been beaten at home. The more traditional supporter may prefer the use of alcohol to relieve stress on these occasions, but we would have to say that its effects on sudden cardiac death remain anecdotal by comparison with aspirin.

Our results also remind us that, to many, football is much more “a matter of life and death” than it is “only a game”.