The statistical significance level was set at 5% and all calculations were two-tailed. All categorical variables are expressed as absolute values using the chi-square test as appropriate. The trend line selected was a fifth order polynomial to compare patterns of times of death in the ICU and GW using Excel version 2010. SPSS statistical software version 20.0 (IBM Corp., Released 2011. Armonk, NY) was used for all statistical analyses.

Several interesting facts emerge when mortality rates between 2006 and 2012 in the two hospital wards of the study site are examined. Malignant neoplasms remain the most common cause of death in the GW, in line with many previous findings, whereas diseases of the circulatory system are the most common cause of death in the ICU (Figure 5). Deaths occur most frequently on weekends for those in the GW, as opposed to those in the ICU, in which deaths occur most frequently on weekdays, particularly Wednesdays or Thursdays (Figure 4). The number of deaths in the ICU notably increased on a yearly basis despite advances in medical techniques and information (Figure 1).

According to Rocha et al. [3], malignant neoplasms (42.3%) remain the most common cause of death in the GW, whereas in the ICU, the frequency of septicemia is staggeringly high (64.0%) and the frequency of acute myocardial infarction (AMI) is very low (2.8%). No conclusions have been drawn, but given the high number of deaths due to diseases of the circulatory system in the ICU, the following may be surmised as reasons for the aforementioned trends in mortality: first, a higher ratio of patients have cardiovascular complications compared with those with noncardiovascular complications. Second, patients may often develop multiple organ failure as well as increased diversity of disease as a result of an increased length of stay in the ICU, and thus die owing to inevitable comorbidity with cardiovascular diseases.

Reasons for the decreasing trend in the number of deaths in the GW after 2008 in contrast with the increasing number in the ICU may include ICU expansion, increased circulatory and respiratory disease as a result of an increase in nosocomial infections and antibiotic-resistant bacteria, and the increased prevalence of infectious diseases in older patients. However, a multifaceted statistical analysis for yearly trends in the cause of death may reveal a more precise reason for this effect.

As previously mentioned, this is in line with previous research showing that malignant neoplasms are the reason for many deaths in the GW. However, in the ICU, patients have complex causes of death, as they not only have malignant neoplasms but also comorbid circulatory, respiratory, and gastroenterological diseases (Figure 5). Therefore, the results of various studies may differ regarding this topic.

Management of medical personnel affects the mortality rate in a hospital. Another study reported that in-hospital mortality rates are affected by human resource management systems [4]. Moreover, mortality rates increase when resident doctors first begin in-hospital work [5]. At the site of the present study, a monthly analysis showed that the mortality rate in the GW was highest in January or July to September, which were time periods corresponding with January and August staff vacations and redistribution of medical personnel (Figure 2). Meanwhile, no statistical significance (p=0.87) was found between the mortality rate in the months of March to May (ICU, 25.1%; GW, 25.0%) and September to November (ICU, 25.0%; GW, 24.2%), although this appeared to be influenced by the aforementioned management of medical personnel.

Differences in mortality rates according to time of day or day of the week have been researched most commonly for cardiovascular diseases. A causal relationship between the increase in alphasympathetic tone and sudden cardiac deaths in the mornings was demonstrated [6]. Muller et al. analyzed the relationship between occurrence of death and the time of day among 2,203 patients in Massachusetts hospitals over the course of a year and proposed two reasons why the incidence of sudden cardiac death was highest from 7:00 AM to 11:00 AM while being lowest at night. First, most cases of sudden cardiac death are due to myocardial infarctions caused by platelet aggregation at atherosclerotic lesions of coronary vessels, as a result of the increase of plaque rupture and coronary arterial tone from increased blood pressure in the morning. Second, fatal arrhythmias are likely to occur in the mornings owing to electrical instability arising from an increase in sympathetic nervous system activity in the morning [7-9]. Mundigler et al. analyzed the causal relationship between 6-sulfatoxymelatonin (aMT6s), a metabolite of melatonin, and circadian rhythms, and found that urinary aMT6s excretion was reduced in ICU patients, as opposed to GW patients, who maintained regular aMT6s excretion and thus preserved their circadian rhythms [10]. In the present study, the high mortality rate in the early morning hours (04:00–06:00 AM) in the GW may be corroborated by studies such as those previously mentioned (Figure 3). The reason for the lack of such an effect in the ICU may be that endogenous circadian rhythms are changed as the excretion of melatonin from the pineal gland is affected by the use of various sympathomimetic and sedative drugs, as well as the environment of a hospital ward, in which daytime and nighttime differences are unclear. Hourly melatonin secretion levels were not measured in the present study, but future studies may be able to analyze the mortality rate differences between the ICU and GW by investigating the causal relationship between melatonin secretion and circadian rhythms.

The present study showed that in the GW, high mortality rates occurred on weekends, whereas in the ICU, higher mortality rates occurred on Wednesdays and Thursdays (Figure 4). Although not many studies have analyzed inpatient mortality rates based on the day of the week, outpatients and inpatients may still be considered to possess similar circadian rhythms, under the assumption that circadian rhythms are relatively well preserved in GW patients in comparison with ICU patients. This contradicts other studies that stated that outpatient sudden cardiac deaths occur most frequently on Mondays, as such deaths are cardiovascular events arising from the emotional stresses related to the start of a new week [11]. It may thus be the case that despite being in the GW, inpatients still show changes in circadian rhythms in comparison to outpatients, owing to changes in living patterns arising from hospital admittance, unfamiliar hospital environments, and changes in endogenous hormone secretion that result from the diagnoses and treatments of various diseases. Meanwhile, Lopez et al. reported that a high number of outpatient sudden cardiac arrests occurred on Wednesdays, although the causes have not yet been identified [12]. Witte et al. cited geographical factors for this difference [11]. For instance, et al. studied day-of-week differences in a Chinese population admitted to a coronary care unit for AMI and reported that the incidence rate on Sundays was relatively low and that no peak incidence was seen on Mondays [13]. Furthermore, another study on the incidence rate of AMI among a Chinese population showed that the rate was highest on Saturdays, with statistical significance [14]. In a Japanese population, the weekly and monthly changes in terms of aortic dissection incidence showed no clear difference between the days of the week [15]. Thus, race may also be a factor involved in differences in circadian rhythms.

Neumann et al. analyzed the hourly, weekly, and monthly mortality rates of terminal cancer patients, and found that rates between 08:00 and 10:00 hours and 16:00 and 18:00 hours were higher than those in night hours. The causes may be related to an increase in platelet aggregation, hypercoagulability, and catecholamine secretion, which are all related to sudden cardiac death. Weekly and monthly comparisons between mortality rates showed unclear results [16].

The differences in the time-of-day, day-of-week, and monthly patterns of mortality rates in the ICU and GW may be accounted for by the preservation of circadian rhythms in GW patients, due to their having fewer disturbances in endogenous hormones and physiological changes. Meanwhile, the changes in the circadian rhythms of the ICU patients might have been caused by the differences in disease distribution, hospital room environments, and use of various drugs. However, further research is needed to precisely determine the cause of the aforementioned effects.