Acute coronary syndrome on Friday the 13th: a case for re-organising services?

Friday the 13th has long been associated with misfortune, anecdotal evidence suggesting that it coincides with catastrophes and danger. A handful of investigations published in medical journals have investigated this association. However, these studies were limited by small sample sizes, and they did not investigate the possibility of a “lucky day” corresponding to this fateful “unlucky” day.

Traditionally, scientists have dismissed the misfortune associated with Friday the 13th as an urban legend, a myth, or even an old wives’ tale. However, we feel that there are too few published studies of the effects of superstition and Friday the 13th, despite the anecdotal evidence. Recently misreported data concerning the so-called “weekend effect” — increased hospital mortality associated with weekend admissions1 — has led to changes in the junior doctor contract in England that provide more staff coverage to compensate for the alleged effect. Should we make similar provisions for Friday the 13th?

The history books are unclear about the exact origins of the fear of Friday the 13th, or paraskevidekatriaphobia. In Europe, one theory is that humans initially used their ten fingers and two feet to count, and greeted what came after 12 with suspicion.2 Other theories suggest that this was the fateful day when Adam (believed to be the first human) was tempted with the forbidden fruit.2 Finally, a view widespread in the Western world is that Jesus Christ was crucified on a Friday, after 13 people were present at the Last Supper. Regardless of the origins of the fear of this date, it has certainly become entrenched in Western culture that it is unlucky if the 13th day of the month falls on a Friday. In the United States alone, it is estimated that about US$900 million in business and productivity is lost on Friday the 13th.3

A number of studies have examined correlations between this date and health problems. One study found that there are more road traffic deaths on Friday the 13th than on other days of the year;4,5 US data show an avoidance of caesarean deliveries on Friday the 13th and, among the Chinese population, an increase in the number of planned births on their lucky day, the 8th day of the month.6 Bruce and colleagues, in a wide-ranging study of emergency department admissions, found a significantly greater number of patients attended with penetrating trauma on Friday the 13th.7 Other studies have reported a lack of an effect of Friday the 13th on admissions and mortality.8-12 Nevertheless, there seems to be sufficient anecdotal evidence that superstition can affect our hospitals to justify re-thinking how we provide health care on such dates.

In this study, we sought to compare, in a large sample population of cardiovascular patients, the mortality for patients admitted to hospital with an acute coronary syndrome (ACS) on Friday the 13th with that for other dates in the year.

Methods

Data sources

We conducted a retrospective cohort analysis of data from the ACS database of South Wales, United Kingdom (not to be confused with New South Wales, Australia), linked with secondary care data from the Patient Episode Database for Wales (PEDW), primary care data from general practices in Wales, and mortality data from the Office of National Statistics (ONS).

The South Wales ACS registry is a retrospective registry based on data identified in the PEDW database; it includes data for all patients admitted to a South Wales hospital with their first ACS (non-ST elevation myocardial infarction or unstable angina) during 1999–2014. Mortality data were extracted from ONS statistics, including date and cause of death as documented on the official death certificate.

Data linkage

The data were pseudo-anonymised and linked in the Secure Anonymised Information Linkage (SAIL) databank, a national Welsh database that links the widest possible range of person-based data with robust anonymisation procedures.13,14 It contains over 500 million person-based records, and is operated and administered by the Health Information Research Unit, Swansea University, for the purposes of real world health-related research.

Statistical analysis

Analysis was carried out in SPSS 19.0 (IBM) and Excel (Microsoft). Outcome data for mortality (to 5000 days ≈ 13 years) were extracted for all patients included in the analysis. Missing mortality data resulting from loss to follow-up were right-censored. Records were subgrouped by admission day number and name; for example, “13 Friday”, “12 Friday”, “13 Monday”.

Survival was analysed in a Cox proportional hazards regression model, comparing mortality rates for all 216 combinations of day number and name. Hazard ratios for mortality and survival curves were generated. P < 0.05 was deemed statistically significant.

Ethics approval

Ethics approval was not required, as the data were made available through a data access agreement between SAIL and Cardiff University, complying with both parties’ data access policies for undertaking research.

Results

A total of 56 062 patients who had an ACS in South Wales during 1999–2014 were identified. The distribution of patients by admission day number and name is included in online Appendix 1.

For most admission day/date combinations (211 of 216), there was no statistically significant difference in mortality between admissions on those days and those on Friday the 13th (Appendix 2). However, a statistically significant reduction in mortality was identified for five dates: Thursday the 15th (hazard ratio [HR], 0.77; 95% confidence interval [CI], 0.59–0.999), Wednesday the 18th (HR, 0.76; 95% CI, 0.58–0.99), Monday the 28th (HR, 0.76; 95% CI: 0.57–0.99), Monday the 30th (HR: 0.75; 95% CI, 0.57–0.99), and Tuesday the 31st (HR, 0.71; 95% CI, 0.51–0.99).

The survival function curves for these dates and the comparator (Friday the 13th) are shown in the Box, indicating a favourable outcome for patients admitted on these dates compared with Friday the 13th.

Discussion

We found that 13-year mortality rates for patients admitted with their first ACS on “unlucky” Friday the 13th during 1999–2014 were similar to those for most other calendar dates. Surprisingly, however, we also identified five potentially “lucky” days on which mortality rates were significantly lower, by 20–30%.

Our findings suggest that patient mortality is affected by the date on which they are admitted to hospital. Ours is the first study in any country to identify potentially “lucky” admission days, with reduced mortality at up to 13 years of follow-up, which may be relevant to health care authorities in terms of service planning. Additionally, this is the first study to examine the effect of Friday the 13th in cardiovascular medicine.

It is worth noting that all five “lucky” days were weekdays. It is unclear why, but the “weekend mortality” effect perhaps plays a role. Indeed, the “weekend mortality” study by Freemantle and colleagues1 has been frequently cited in medical literature and in the popular media, and has been the basis of the case for expanding 7-day services. In their study, however, Wednesday was the reference weekday, and the weekend was defined as Friday to Monday. This is not consistent with our findings of two lucky Mondays and the absence of a “weekend mortality” effect. Nevertheless, our findings should be taken into account in future junior doctor contract negotiations, and may support a case for reducing services or staffing levels on the “lucky days”, given the favourable outcomes that can only be explained by a supernatural phenomenon.

The strengths of our study include our using data for large number of patients from real-life national records, reducing the likelihood of selection bias. In addition, by using the SAIL database, we were able to determine patient-specific mortality outcomes with anonymised indices.

The limitations of our study are related to the selective nature of the subgroup analysis. It has been reported in the literature that this approach reduces statistical power, increases variance, and causes overinterpretation of what has been described as a “play of chance”.15 This happens because, even with statistical significance defined as P < 0.05, there is a 1 in 20 chance that a “significant” result is a false positive. Consequently, by dividing our data into 217 subgroups we would expect as many as ten statistically significant results. This has previously been described and observed in the subgroup analysis of the landmark cardiovascular trial, the “International Study of Infarct Survival” (ISIS-2), in which the large, well randomised population was divided into 12 astrological star sign groups, two of which proved to be associated with adverse effects linked to aspirin use.15,16 Clearly, therefore, our findings need to be interpreted with extreme caution, given this inflation of the type I error.

Conclusion

For most calendar days, there was no difference in mortality rates for patients admitted with their first ACS from that of “unlucky” Friday the 13th. However, patients who were admitted on Thursday the 15th, Wednesday the 18th, Monday the 28th, Monday the 30th or Tuesday the 31st were 20–30% more likely to survive at least 13 years, suggesting that these dates are potentially “lucky” dates. Despite this finding, we would advise against delayed hospital presentation for patients who experience chest pain on other dates.

Box – Survival function curves for patients admitted with their first acute coronary syndrome on Friday the 13th or on one of five “lucky days”