Friday Leukemia—a Structural Phenomenon

Despite the recent focus on evidence-based medicine, a number of clinical myths live on. The subjective assessment of work-related stress levels in emergency care, where no planning of workload is possible, is especially susceptible to superstition. Although, fortunately, there are now randomized data to show the harmlessness of describing a shift in the emergency room as “quiet” (1), in hematology the conviction persists that acute leukemia is diagnosed particularly often on Friday afternoons (2). Proponents of this hypothesis can even point to data from researchers in Aachen who showed a bias towards initial diagnosis on a Friday among 197 patients (2). In the absence of validation in a second cohort, however, it remains unclear whether this was a robust observation. Furthermore, a research group in Mexico was unable to confirm the results in the context of the healthcare system there (3).

Methods

We investigated 1520 patients included in the AMLCG1999 study (n = 866, recruited 1999–2007), the AMLCG2008 study (n = 274, recruited 2009–2012), or the AMLCG registry (n = 379, recruited 2011–2019) who received intensive induction chemotherapy for the treatment of acute myeloid leukemia (AML). The patients (median age 58 years [y], quartiles 46 y and 66 y) were classified according to the day of initial diagnosis of AML to establish whether Friday leukemia really exists. Overlapping patient data were excluded. Cytomorphological confirmation of all diagnoses was obtained by examination of bone marrow and peripheral blood smears (4, 5).

To answer our primary research question, the association between diagnosis and weekday—excluding Saturday and Sunday—was examined using the two-sided chi-square test. The level of significance was defined as 5%. In further exploratory analyses we determined the correlation of clinical characteristics with Friday leukemia by means of Fisher‘s exact test. Time-to-event analyses were conducted according to the Kaplan–Meier method and compared using the log-rank test. We validated our results against a cohort of 634 unselected patients with acute lymphoblastic leukemia (ALL) (median age 48 y, quartiles 33 y and 65 y) who received their initial diagnosis or confirmation of recurrence in the same period as the patients with AML (1999–2019). All diagnoses were made by the staff of the Laboratory for Leukemia Diagnosis at LMU Munich.

Results

There were indeed significantly more diagnoses on Fridays than there were per day on Mondays to Thursdays (AML: 326 (23%) versus 1083 (77%), p = 0.006; ALL: 135 (24%) versus 424 (76%), p = 0.019 (Figure). We looked for biological peculiarities of the Friday leukemias in the AML study patients. A Kaplan–Meier analysis showed no difference between Friday and other days of the week with regard to overall survival (AML: p = 0.8). There was still no difference after exclusion of initial diagnoses at the weekend (data not shown; AML: p = 0.8). The frequency of early death (up to 30 days after diagnosis) was not higher (Table). No association was found between AML diagnosed on a Friday and classification in the 2017 European LeukemiaNet risk stratification system (ELN 2017). Therewas also no obvious relation to the sex of the patient. The leukocyte and platelet counts and hemoglobin level at first diagnosis showed no difference. This was also the case for lactate dehydrogenase (LDH).

Figure.

Initial diagnoses of AMLCG study patients by day of the week

The distribution of initial diagnoses over the days of the week in the AML cohort. With weekend diagnoses excluded, there were significantly more initial diagnoses of AML on Fridays than on the remaining weekdays. AML, Acute myeloid leukemia

Table. Characteristics of patients with AML.

AMLCG study patients	Non-Friday	Friday	p
n	1194	326	–
Age in years (quartiles)	58 (46–66)	60 (45–66)	0.35
Female sex, n (%)	567 (47%)	173 (53%)	0.16
ELN risk stratification
Favorable, n (%)	445 (37%)	116 (36%)	0.88
Intermediate, n (%)	283 (24%)	81 (25%)
Adverse, n (%)	406 (34%)	109 (33%)
Unavailable, n (%)	60 (5%)	20 (6%)	–
Laboratory
Leukocytes (quartiles) (G/L)	17.4 (4–55.4)	19.7 (4.9–51.1)	0.35
Hb (quartiles) (g/L)	9 (7.9–10.3)	8.8 (7.9–9.9)	0.39
Platelets (quartiles) (G/L)	55 (29–101)	59 (35–103)	0.95
LDH (quartiles) (U/L)	436 (268–771)	432 (294–758)	0.87
Survival
Early deaths, n (%)	91 (8%)	20 (6%)	0.4
Median overall survival in years (95% confidence interval)	1.6 [0; 17.6]	1.7 [0; 17.1]	0.8

AML, Acute myeloid leukemia; Hb, hemoglobin; LDH, lactate dehydrogenase

Discussion

It must be concluded that Friday leukemia exists, at least in the German healthcare system. Our cohorts, cumulatively representing a period of 20 years and two entities, show that the effect is reproducible. In agreement with Wilop et al. (2), we suspect that the biological characteristics of this entity probably play a minor role. If elevated aggressiveness of the disease were prompting the patients concerned to present before the onset of the weekend, then higher average LDH, a higher early death rate, and perhaps also poorer ELN 2017 risk stratification and worse overall survival would be expected. However, all of these were absent, so no greater severity of the Friday cases can be assumed. Rather, our findings should prompt efforts to identify reasons for this imbalance embedded in the healthcare system. Since the diagnoses were made in our diagnostic laboratory and had high clinical priority, entry of an incorrect later date of diagnosis is unlikely. Anecdotally we have heard of numerous patients who had to visit a succession of doctors (typically: primary-care physician, office-based hematologist, small hospital, hematology center) before their diagnosis was confirmed and treatment initiated. This long pathway may—together with the understandable desire of all involved to get a potentially dangerously ill patient started on appropriate treatment with the weekend looming—partially explain the accumulation of Friday diagnoses. The trend to fewer Monday diagnoses, discernable in the Figure, is in line with this explanation. A patient who sees a doctor on a Monday will often not arrive at the diagnosing center until a few days later. We are therefore of the opinion that greater awareness of the diagnosis and treatment of the disease is required. A simple differential blood count in the outpatient setting can harden the suspicion of acute leukemia by detecting precursor cells. Moreover, suspicion of leukemia should (also according to the guidelines of the German Society of Hematology and Oncology [DGHO]) prompt direct referral to a center for hematology and oncology, because treatment in the framework of a study is very important for these diseases. This and similar measures might lead not only to reduction of the stressful time before patients receive their diagnosis but also to more efficient use of healthcare resources.