A bad time to get sick?

“NHS hospital deaths rise on [the] day junior doctors join wards,” The Guardian reported. It said the NHS has its own annual “black Wednesday”, when deaths go up by an average of 6% the day newly qualified doctors start work.

The report is based on a large retrospective analysis of hospital records. It compared the number of deaths in emergency cases on the last Wednesday in July and the first Wednesday in August. It found that patients admitted on the first Wednesday in August (the week that new doctors traditionally start working), were 6% more likely to die than those admitted the week before.

Although a statistically significant difference in the number of deaths was found, the absolute difference was small (45 patients out of almost 300,000 records for nine years). The exact causes of death are unknown, and unless individual patient records are examined, whether the deaths were preventable. However, the timing coincides with the period that junior doctors begin their first ward rounds.

The authors of the study say that the mortality difference is “small but significant” and that this study could not explore the specific factors influencing the clinical outcomes. They call for more research measuring preventable deaths as a marker of the effect of the junior doctor changeover.

Where did the story come from?

This study was carried out by Dr Min Jen and colleagues from the Dr Foster Unit and other academic departments at Imperial College in London. The Dr Foster Unit is funded by a grant from Dr Foster Intelligence, an independent health service research organisation, and has indirect support from the National Institute of Health Research.

The study was published in the peer-reviewed medical journal PLoS ONE .

What kind of scientific study was this?

The aim of this retrospective cohort study was to compare survival between a group of patients admitted as an emergency on the last Wednesday in July and the first Wednesday in August. The August date corresponds with a week that hospitals typically take on trainee doctors.

The researchers reviewed hospital records from the Hospital Episode Statistics database for patients admitted into hospital between 2000 and 2008. Each of the two groups of patients was followed up for a week after their admission on either the week in July or August, and deaths during that week were recorded. Only hospitals that take on trainee doctors on the first Wednesday in August each year were included.

The odds of dying in the week following the first Wednesday in August were compared with the odds of dying in the week following the last Wednesday in July. The calculations adjusted for factors that may have affected the risk of death, including age, gender, socioeconomic status, other illnesses, as well as the year and the week of diagnosis. In a separate analysis, the researchers looked specifically at the effects of the changeover in 2007 and 2008. In 2007, there was a controversy over the Medical Training Application Service (MTAS), the NHS’ recruitment service.

What were the results of the study?

A total of 299,741 patients were admitted on these two days in the years from 2000 to 2008. Of those, 151,844 were admitted on the last Wednesday in July and 147,897 on the first Wednesday in August. In total, there were 4,409 deaths in the two groups, 2,182 among those patients admitted on the last Wednesday in July and 2,227 among those patients admitted the week after.

The researchers noted that throughout the nine-year period, there were fewer admissions each year on the first Wednesday in August than on the Wednesday in July. There were little other differences between the groups.

When the researchers adjusted for potential confounding factors, they found that the odds of death in the group admitted in August was 6% higher than the group admitted in July (odds ratio [OR] 1.06, 95% confidence interval [CI] 1.00 to 1.15, p = 0.05). The pattern was consistent across the years, except when the researchers analysed 2007 and 2008 separately (where there was no statistically significantly greater odds of dying in August).

When the analyses were divided by the reasons for admission, those admitted for surgery (12.3% of all admissions) or for cancer (2.8% of all admissions) did not have significantly different survival rates between the two weeks. However, the majority of admissions (85%) were for patients with a diagnosis in the medical category (e.g. heart attacks or strokes), and this group was 8% more likely to die if they were admitted on the Wednesday in August than the Wednesday in July.

What interpretations did the researchers draw from these results?

The researchers conclude that in a large study of emergency admissions, there was a small but significant 6% higher odds of death for all patients in the week following the first Wednesday in August than in the week following the last Wednesday in July. The August admissions coincide with the time when all UK hospitals “plan and undertake the changeover of junior doctors”.

What does the NHS Knowledge Service make of this study?

This is a large, generally well-conducted retrospective cohort study. It noted a small but significant difference in mortality for emergency hospital admissions on the first Wednesday in August compared with the last Wednesday in July.

This observation, referred to in the US as the 'July phenomenon', has been investigated in several studies with inconsistent findings. In this study, the researchers attempted to avoid the possible biases that the other studies investigating this July phenomenon may have faced. For example:

• They only included emergency admissions to address the possible bias that elective admissions may be planned around the holiday period and upcoming change in hospital staff.
• They followed patients for a short period (one week) because later deaths may not reflect the quality of initial management and care.
• They used routinely collected data for a large number of patients.

There are some points to bear in mind when considering the findings of this study:

• The absolute number of increased deaths in the August admissions was small, that is, a total of 45 in nine years (2,227 compared with 2,182). This is 45 deaths too many, particularly if there is a link to the quality of care these patients received, but the 6% increased odds must be interpreted alongside the actual numbers who died.
• The research highlighted a consistent pattern across the years: that admissions on the first Wednesday in August were consistently fewer than the number of admissions on the last Wednesday in July. This pattern suggests some systematic differences in referrals that needs further investigation. For example, patients with less severe illness, and who have the choice, may prefer to be admitted in July before the staff handover. As such, unmeasured factors in relation to who is admitted and why they are admitted may play a part.
• With data on almost 300,000 admissions, the results are statistically significant but only just.

From these results, it isn’t possible to conclude that poorer care is responsible for the higher deaths in the first week in August. Further study of individual cases and a quantification of the number of preventable deaths would be helpful and the researchers specifically call for this.