Predictors of in-hospital mortality in nonsurgical departments: a multivariable regression analysis of 2 855 029 hospitalizations

ABSTRACT

Introduction: In-hospital mortality is a relevant outcome of hospital admissions.

Objectives: This study aimed to identify predictors independently associated with in-hospital mortality in nonsurgical departments.

Patients and methods: In 2014, the Polish National Health Fund database provided data on 2855029 hospitalizations of adults, which were not related with surgical procedures. Patients’ age and sex, diagnosis-related group category assigned to the hospitalization, length of stay, types of hospital and admission, and day of the week and month of admission were analyzed as mortality predictors.

Results: The mean (SD) in-hospital mortality rate was 4.1% (0.01%). Odds ratios for in-hospital death increased with patients’ age. The female sex was associated with lower odds ratios of death than the male sex. Among the diagnosis-related groups assigned to hospitalizations, the highest mortality was found in patients with vascular diseases (11.95%). Considering the length of stay, the lowest mortality occurred during 5- to 7-day hospitalizations (2.63%). Compared with teaching hospitals, the odds ratio of death was 1.31-fold higher for regional hospitals, 1.35-fold higher for private hospitals, and 1.48-fold higher for district and city hospitals; 92% of all in-hospital deaths occurred after urgent and emergency admissions. Hospital admissions at weekends or on other nonworking days (public holidays) were significant predictors of in-hospital mortality. Differences in mortality rates were found between particular months, but no seasonal relationship was established.

Conclusions: Age, male sex, emergency admission, admission at the weekend or on another nonworking day (during public holidays), and hospitalization in a district, city, private, or regional hospital (compared with a university hospital) were factors associated with higher mortality in nonsurgical departments.

What’s new?

A better understanding of hospital mortality predictors can improve hospitalization outcomes. This is the first study aimed to identify the predictors independently associated with in-hospital mortality in a large population of patients hospitalized in nonsurgical departments. Aware of such predictors, doctors can identify patients with higher mortality risk and take steps to improve healthcare quality. The admission day was one of predictors evaluated in our study. Previous studies identified the “weekend effect,” suggesting that patients who are admitted to hospital at the weekend have increased mortality risk. We showed higher mortality in patients admitted not only during weekends, but also on Wednesdays, Thursdays, Fridays, and public holidays, which has not been reported so far.

Introduction

In-hospital mortality (deaths that occur during hospitalization) is a relevant outcome of hospital admission and, like other measures of mortality (eg, 30-day mortality), has been used to assess the quality of care. Significant variations in hospital mortality rates are commonly reported and remain unclear to a great extent.^1,2 Knowledge of in-hospital mortality predictors is crucial for improving patients’ outcomes. A better understanding of in-hospital mortality predictors may help to identify high-risk patients and develop strategies to reduce mortality rates.

In-hospital mortality depends on numerous factors related to patients, medications,³ and even environmental conditions⁴ and varies among hospital departments. No studies evaluating in-hospital mortality in a large population of patients hospitalized in nonsurgical departments have been conducted so far. In those evaluating general in-hospital mortality, it was demonstrated that age, sex, diagnosis, comorbidities, mode of admission (urgent vs elective), need for transfer between hospitals, number of previous emergency admissions, and length of hospital stay were the most relevant patient-related factors for in-hospital mortality. Other factors, such as healthcare system organization and financing, hospital type, hospital volume, use of safety checklists, percentage of individuals admitted to the critical care unit, ratio of hospital doctors to the number of beds, and ratio of general practitioners to the population size, may also matter.^2,5-8

The number of people aged over 65 years is significantly and rapidly growing, and the elderly account for a high percentage of patients who die in hospitals. Factors specifically associated with in-hospital mortality in the elderly can be determined, for example, functional and cognitive impairment, male sex, disease severity, comorbidity scores, polypharmacy, abnormal levels of parameters routinely checked on admission, emergency events, and hospital admission.^9-12

As mentioned above, although in-hospital mortality is a significant measure of healthcare quality, the issue has not been fully covered in the literature, the populations studied were relatively small,^13,14 and the articles focused mostly on deaths related to specific causes, groups of diseases, or particular departments.^15-17 Therefore, this study aimed to identify predictors that are independently associated with in-hospital mortality in a large population of patients hospitalized in nonsurgical departments.

Patients and methods

Data from the year 2014 were obtained from the database of the Polish National Health Fund (Narodowy Fundusz Zdrowia [NFZ]), a public organization financing medical procedures in Poland. Almost all Polish citizens are insured by NFZ.¹⁸ All healthcare services, if supposed to be paid for by NFZ, have to be contracted with this institution. Healthcare providers keep an electronic registry of healthcare services and collect data on each service. In all hospitals in the country, entries have to be made on the day of hospital discharge or patient’s death. The data that should be transferred to NFZ include, among others, patients’ name, personal ID number, sex, age, hospitalization dates, type of admission (elective, urgent, or emergency), discharge status (including death), diagnosis according to the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision, DRG number,¹⁸ and major diagnostic category (nonsurgical procedures are classified according to the NFZ system, that is, based on DRG sections: nervous system, gastrointestinal tract, liver, pancreas, and spleen, heart and circulation, vascular system, respiratory system, endocrine system, musculoskeletal system, hematopoietic system and intoxication and infectious diseases, head and neck, skin and mammary gland, genitourinary tract and female genital tract).

This study included data regarding only nonsurgical hospitalizations of adult patients (aged 18 years and older). Hospitalizations classified as “other” were excluded.

“In-hospital death” was defined as death during hospitalization. Such cases were identified as a discharge status predefined in the NFZ database as “death.” Death after hospital discharge was not recorded. A crude mortality rate was expressed as the number of hospitalizations that ended with death in a given year, divided by the total number of hospitalizations in that year.

The following mortality predictors were included in the analysis: patients’ age (patients were divided into 9 age groups: 18–24, 25–34, 35–44, 45–54, 55–64, 65–74, 75–84, 85–94, and ≥95 years) and sex, DRGs assigned to hospitalizations, the length of hospital stay, hospital type (regional, district, city, teaching, or private), admission type (elective admission: scheduled in-hospital stay following physician’s referral; urgent admission: in patients whose condition required quick but not immediate action; or emergency admission: in patients transferred to the hospital by ambulance and whose condition required immediate action), month of admission, day of the week on which admission took place, and type of an admission day, that is, a regular weekday or a holiday.

Statistical analysis

The multivariable logistic regression model including all the aforementioned predictors was used to estimate the adjusted odds ratios (ORs). To handle the quasi-complete separation of data, the Firth correction was applied. The following tests were performed to assess model adequacy (a P value less than 0.05 was considered significant): the likelihood ratio test for the joint significance of predictors, the Wald χ² test for single-predictor significance, as well as the Hosmer–Lemeshow and Osius–Rojek goodness-of-fit tests.

Odds ratios greater than 1 indicated a higher risk of in-hospital death, and these below 1, a lower risk of in-hospital death in the analyzed group compared with the reference group. Furthermore, 95% CIs were calculated for ORs. The Wald χ² test was performed to verify whether ORs were different than 1. The analysis was conducted using the SAS system (SAS Institute Inc., Cary, North Carolina, United States).¹⁷

Results

In 2014, there were 2855029 hospitalizations (of 1634104 women and 1220925 men) and 116971 deaths in nonsurgical departments. The mean (SD) in-hospital mortality rate was 4.1% (0.01%). There were 59135 deaths in women (3.62%) and 57836 in men (4.74%). The female sex was associated with a lower mortality risk than the male sex (fully adjusted OR, 0.79; P <0.001).

The mortality rates differed among age groups and increased with age: 160 deaths (0.1%) were reported in patients aged 18 to 24 years, 21859 (4.15%) in those aged 65 to 74 years, 37943 (6.99%) in those aged 75 to 84 years, 31338 (14.04%) in those aged 85 to 94 years, and 2583 (25.33%) in those aged 95 or older. The fully adjusted OR for in-hospital death increased with patients’ age, being 198.2-fold higher in patients aged 95 years or older than in those aged 18 to 24 years. Fully adjusted ORs for in-hospital death in nonsurgical departments for particular age groups are shown in Table 1.

**Table 1.** In-hospital mortality in nonsurgical departments stratified by the analyzed mortality predictors
Risk factor		DF	Estimate	SE	Wald χ²	P value	OR	95% CI
Intercept		1	–6.24	0.08	5596.66	<0.001	–	–
Female sex		1	–0.23	0.01	1249.48	<0.001	0.79	0.78–0.8
Age^a, y	25–34	1	0.77	0.09	73.74	<0.001	2.15	1.81–2.57
	35–44	1	1.77	0.08	452.04	<0.001	5.85	4.97–6.88
	45–54	1	2.58	0.08	1022.22	<0.001	13.23	11.3–15.15
	55–64	1	3.1	0.08	1509.87	<0.001	22.29	19.06–26.07
	65–74	1	3.48	0.08	1900.61	<0.001	32.43	27.73–37.92
	75–84	1	3.99	0.08	2500.99	<0.001	53.86	46.07–62.96
	85–94	1	4.66	0.08	3409.19	<0.001	105.69	90.38–123.6
	≥95	1	5.29	0.08	4025.07	<0.001	198.18	168.3–233.4
Admission type^b	Urgent	1	1.52	0.01	16 009.08	<0.001	4.57	4.45–4.67
Admission type^b	Emergency	1	2.22	0.01	30 085.42	<0.001	9.16	8.94–9.4
Hospital type^c	District and city	1	0.39	0.01	1087.93	<0.001	1.48	1.45–1.52
	Private	1	0.3	0.01	426.36	<0.001	1.35	1.31–1.39
	Regional	1	0.27	0.01	473.74	<0.001	1.31	1.28–1.34
Admission day^d	Tuesday	1	0.02	0.01	3.02	0.08	1.02	1–1.04
	Wednesday	1	0.1	0.01	68.16	<0.001	1.1	1.08–1.12
	Thursday	1	0.12	0.01	107.32	<0.001	1.13	1.1–1.15
	Friday	1	0.25	0.01	478.56	<0.001	1.29	1.26–1.31
	Saturday	1	0.33	0.01	712.57	<0.001	1.4	1.36–1.43
	Sunday	1	0.25	0.01	412.23	<0.001	1.29	1.26–1.32
Public holiday^e		1	0.22	0.02	116.73	<0.001	1.25	1.2–1.3
Admission month^f	February	1	0.01	0.02	0.4	0.53	1.01	0.98–1.04
	March	1	0	0.02	0.01	0.93	1	0.97–1.03
	April	1	–0.03	0.02	2.68	0.1	0.98	0.95–1
	May	1	–0.08	0.02	29.22	<0.001	0.92	0.89–0.95
	June	1	–0.1	0.02	36.02	<0.001	0.91	0.88–0.94
	July	1	0.02	0.02	1.5	0.22	1.02	0.99–1.05
	August	1	–0.01	0.02	0.64	0.42	0.99	0.96–1.02
	September	1	–0.03	0.02	3.46	0.06	0.97	0.94–1
	October	1	0.02	0.02	1.72	0.19	1.02	0.99–1.05
	November	1	0.02	0.02	1.29	0.26	1.02	0.99–1.05
	December	1	0.05	0.02	9.22	0.002	1.05	1.02–1.08
Length of stay, d	1	1	0.18	0.01	156.52	<0.001	1.2	1.17–1.24
	2–4	1	–1.53	0.01	12 838.36	<0.001	0.22	0.21–0.22
	5–7	1	–2.13	0.01	21 849.93	<0.001	0.12	0.12–0.12
	8–14	1	–1.96	0.01	19 308.49	<0.001	0.14	0.14–0.15
	15–21	1	–1.2	0.02	5307.98	<0.001	0.3	0.29–0.31
	22–30	1	–0.61	0.02	979.23	<0.001	0.54	0.52–0.56
	31–60	1	–0.09	0.02	17.39	<0.001	0.91	0.88–0.95
	≥61	1	0.48	0.04	128.58	<0.001	1.62	1.49–1.76
a Compared with data for age (range), 18–25 years b Compared with data for elective admission c Compared with data for teaching hospitals d Compared with data for Monday e Compared with data for working days f Compared with data for January Abbreviations: DF, degrees of freedom; OR, odds ratio

The highest mortality rate was associated with vascular diseases: as many as 6758 deaths (11.95%) were reported in patients with these conditions. The next 3 groups of diseases with the highest mortality rates were pulmonary diseases with 23433 deaths (7.30%), liver diseases with 7669 deaths (5.87%), and heart diseases with 29509 deaths (5.37%).

In 2014, there were 1479759 urgent, 391727 emergency, and 983543 elective admissions with the mean (SD) in-hospital mortality rates of 4.23% (0.02%), 11.5% (0.05%), and 0.95% (0.01%), respectively. Of all in-hospital deaths, 107648 (92%) followed urgent and emergency admissions. Compared with elective admissions, the mortality risk was 4.56-fold higher for urgent admissions and 9.16-fold higher for emergency admissions (P <0.001). Fully adjusted ORs for in-hospital mortality in nonsurgical departments for particular admission types are shown in Table 1.

In 2014, a total of 457435 hospitalizations took place in teaching hospitals, 1267234 in district and city hospitals, 831191 in regional hospitals, and 299169 in private hospitals. The mean (SD) in-hospital mortality rate was 2.18% (0.02%) in teaching hospitals, 4.72% (0.02%) in district and city hospitals, 4.09% (0.02%) in regional hospitals, and 4.44% (0.04%) in private hospitals. Fully adjusted ORs for in-hospital mortality in nonsurgical departments for particular hospital types are shown in Table 1.

In 2014, there were 476150 admissions at weekends or on other nonworking days (the mean [SD] in-hospital mortality in this patient group was 6.49% [0.04%]). The daily number of hospitalizations on nonworking days was about 2-fold lower than that on working days (eg, 200486 hospitalizations on Saturdays vs 566597 hospitalizations on Mondays). Hospital admission at weekends or on other nonworking days (public holidays) was a significant predictor of in-hospital mortality. As compared with Mondays, the number of deaths was also higher on Wednesdays, Thursdays, and Fridays. Fully adjusted ORs for in-hospital mortality by days of admission are shown in Table 1.

The smallest number of admissions (218700) was recorded in December, and the largest (258770) in March. The highest mortality was noted in December—10 334 deaths (4.73%), and the lowest in June—8725 deaths (3.76%). As compared with January, mortality was lower in May (OR, 0.92; P <0.001) and June (OR, 0.91; P <0.001), but higher in December (OR, 1.05; P = 0.002). Fully adjusted ORs for in-hospital mortality by months of admission are presented in Table 1.

In-hospital mortality was also associated with the length of hospital stay. The lowest mortality was observed during hospitalizations lasting for 5 to 7 days (17820 deaths [2.63%]). The highest mortality was observed during very short hospital stay (10322 deaths [4.71%] for less than 1 day of stay and 17245 deaths [9.39%] for 1 day of stay) and very long hospital stay (1006 deaths [26.49%] for stay longer than than 60 days) with mortality rising gradually after day 8. Fully adjusted ORs for in-hospital mortality in nonsurgical departments by the length of hospital stay are presented in Table 1.

Among hospitalizations lasting 1 day or shorter, the elective ones predominated, and the in-hospital mortality in this category was low. The high in-hospital mortality was observed in patients hospitalized for 1 day or shorter, who were admitted urgently (especially on an emergency basis). The in-hospital mortality for particular hospitalization time intervals and particular types of admission is shown in Table 2.

**Table 2.** In-hospital mortality in nonsurgical departments by length of hospital stay and admission type
Length of hospital stay, d	Admission type	Hospitalizations, n	Deaths, n	Mortality rate, %
0	Elective	155164	454	0.3
	Urgent	55838	4968	8.9
	Emergency	12551	5153	41.1
1	Elective	73952	973	1.3
	Urgent	89495	8984	10
	Emergency	24353	7695	31.6
2–4	Elective	391161	1995	0.5
	Urgent	539000	15119	2.8
	Emergency	111589	11354	10.2
5–7	Elective	205951	1564	0.8
	Urgent	392474	9915	2.5
	Emergency	99343	6761	6.8
8–14	Elective	156418	2110	1.3
	Urgent	331399	12834	3.9
	Emergency	109946	8056	7.3
15–21	Elective	24996	1035	4.1
	Urgent	66961	5680	8.5
	Emergency	26299	3319	12.6
22–30	Elective	7542	677	9
	Urgent	23831	3213	13.5
	Emergency	10258	1967	19.2
31–60	Elective	3622	548	15.1
	Urgent	13112	2630	20.1
	Emergency	6639	1586	23.9
≥61	Elective	545	137	25.1
	Urgent	2142	525	24.5
	Emergency	1198	371	31

Discussion

In this study, we analyzed predictors of in-hospital mortality in nonsurgical departments, using a comprehensive database on hospitalizations in Poland. To the best of our knowledge, this is the first analysis of a large national database to assess general (not associated with any specific disease or medical specialty) in-hospital mortality, not related with surgery. It may help predict mortality in patients hospitalized in nonsurgical departments more accurately. We found that advanced age, male sex, urgent and emergency admission, admissions at weekends or on other nonworking days (public holidays), and hospitalization in a district, city, private, or regional hospital (vs teaching hospital) were factors associated with greater mortality.

On one hand, the prognostic significance of sex in hospitalized patients in the general population has not been extensively studied. On the other hand, some papers on this subject were published in the field of cardiology.^20-24 In most of these studies, female sex was associated with higher in-hospital mortality. One study evaluating in-hospital mortality after stroke showed that sex was not an independent predictor of death.²⁵ However, another study presented significant differences in in-hospital mortality with regard to sex and demonstrated that mortality in men of working age was 2-fold higher than in women from the same age group.²⁶ We found only 2 studies comparing in-hospital mortality rates between men and women in the population of patients hospitalized in general nonsurgical departments. In the study by Hwang et al,¹²which included 2867 patients in a large urban university hospital in Taiwan, who were older than 75 years, a negative association between the female sex and in-hospital mortality was observed in an adjusted regression model. In the second study by Gordon et al,²⁷which included 89793 patients hospitalized in 30 hospitals in Northeast Ohio and diagnosed with 6 common conditions not requiring surgery, the in-hospital death rates, adjusted for disease severity, were lower in women than in men. In a report on in-hospital mortality in surgical and nonsurgical departments, which was prepared by the Mississippi State Department of Health and based on hospital discharge data from all the reporting hospitals in Mississippi in 2010 (including patients of all ages), the in-hospital mortality rate was 2.1% in women, and 2.7% in men.²⁸ In a similar analysis conducted in Canada,⁵ men were more likely to die in the hospital than women (OR, 1.09). In our study, the mean (SD) in-hospital mortality rate was 3.62% (0.01%) in women, and 4.74% (0.02%) in men. Therefore, the female sex was associated with a lower risk of death than the male sex (OR, 0.79; 95% CI, 0.78–0.8). Importantly, we analyzed only nonsurgical departments, adult patients, and almost all hospitalizations of this kind that occurred in the whole country during 1 year.

In our study, mortality was highly dependent on age, which is consistent with the observations of other authors.^5,28,29 It is not surprising, as older patients are more likely to have multiple pre-existing conditions and, usually, are in worse condition than younger ones.

The in-hospital mortality in nonsurgical departments is higher in patients admitted urgently. In the years 2014 to 2015, more than a half of deaths (58%) in Australian hospitals (in patients of all ages in both surgical and nonsurgical departments) were attributed to emergency admissions, whereas 21% were associated with elective admissions (the urgent status was not assigned or reported for 21% of deaths).³⁰ In hospitals in the state of Mississippi, United States, emergency hospital admissions accounted for 55.4% of in-hospital deaths.²⁸ In our study, 92% of in-hospital deaths were associated with urgent and emergency admissions. The risk of death was 4.56-fold higher for urgent admissions and 9.16-fold higher for emergency admissions compared with the elective ones. The analysis performed by the Canadian Institute for Health Information, which concerned hospital mortality trends in Canada and included patients admitted to acute care hospitals outside Quebec between April 2004 and March 2005, showed that only 14% of patients who died in the hospital were admitted electively.⁵ The OR of death was 2.6-fold higher in their counterparts admitted on an urgent or emergency basis. In turn, a prospective 1-year cohort study conducted in the Internal Medicine Hospitalization Unit at the Hospital in Lima (Peru) evaluated 499 adult patients with type 2 diabetes and found no association between the type of hospital admission (outpatient service or emergency) and in-hospital mortality.³¹ In another study examining the determinants of in-hospital mortality in adult patients with heart failure, which was a retrospective observational study using the 2010 Nebraska Hospital Discharge data on 4319 hospitalizations and 79 hospitals in the state of Nebraska, United States, the type of admission was not considered a risk factor for in-hospital death.³²

The relationship between hospital mortality and a hospital type (for example, teaching or private) received considerable attention in the literature. In the study on a database of all admissions to acute care hospitals in the state of California, United States, in 1998, compared with hospitals other than university hospitals, mortality was similar in patients admitted to major and minor university hospitals.³³ In turn, a recently published study on 21.5 million hospitalizations of the Medicare beneficiaries at 4483 hospitals across the United States between 2012 and 2014 showed that, compared with hospitals other than university hospitals, lower mortality rates for common conditions was reported in major university hospitals.³⁴ In our study, teaching hospitals (vs district and city, regional, and private hospitals) were also characterized by the lowest in-hospital mortality rate. An analysis of 16205314 discharges in the years 2001 to 2010 in Chile³⁵ showed a lower risk of in-hospital mortality in private hospitals compared with public ones. In our study, the in-hospital nonsurgical mortality in private hospitals was higher than that in regional public hospitals and university hospitals but lower than that in district and city public hospitals.

Previous studies identified the “weekend effect,” meaning that patients who are admitted to hospital at weekends have an increased mortality risk.^36-39 This may be due to a lower number of hospital employees working on those days, less experienced staff, longer time waiting to receive treatment, limited access to test results and diagnostic modalities, and serious condition of patients (only the sickest patients are admitted at weekends). However, not all studies showed increased in-hospital mortality in patients admitted to the hospital at weekends.^40,41 In our study, hospital admissions at weekends, as well as those on other nonworking days (public holidays), were significant predictors of in-hospital mortality. We also observed that the number of deaths was higher on Wednesdays, Thursdays, and Fridays than on Mondays.

We found only a few studies assessing seasonal variation in in-hospital mortality and all these articles focused on deaths related to specific groups of diseases. In Japanese patients hospitalized with Takotsubo syndrome, in-hospital mortality varied widely between months, from 3% in September to 7.5% in April.⁴² A study on the National Inpatient Database 2005–2014 in the United States showed that in patients with a primary diagnosis of bowel ischemia, the highest in-hospital mortality was reported in January and December, and the lowest in July and September.⁴³ In patients hospitalized in the Department of Cardiology in Spain, mortality ranged from 0.17 deaths per day in August to 0.4 deaths per day in February.⁴⁴ In our study, we found no seasonal variation in in-hospital nonsurgical mortality; however, differences between particular months were seen: the highest in-hospital mortality was noted in December (4.73%), and the lowest in June (3.76%). Compared with January, mortality was significantly lower in May (OR, 0.92) and June (OR, 0.91), yet higher in December (OR, 1.05). These differences are difficult to explain and do not seem to result from employing new young residents or be related to the number of hospitalized patients.

In-hospital mortality is also associated with the length of hospital stay. In the United States and Australia, patients who eventually died in the hospital stayed in there longer than other patients.^30,45 In Canadian patients, higher in-hospital mortality was recorded for short (lasting 1 day or 2 days; OR, 3.7 and 1.8, respectively) and long (lasting 22 to 365 days; OR, 1.53) stay compared with that of 3 to 9 days. This is consistent with the results of our study: in nonsurgical departments, we observed the highest mortality for hospitalizations lasting 1 day or shorter (and those longer than 60 days), and the lowest for those lasting 5 to 7 days. The higher OR for hospitalizations lasting 1 day or shorter (hospitalizing patients for 1 day is slightly risky) was possibly the result of high mortality in patients requiring urgent or emergency admission. In those patients, the mortality rate for emergency admissions was 137-fold higher than for elective admissions, whereas only a 22-fold higher difference was noted in the group hospitalized for 5 to 7 days. However, relationships between the length of hospital stay and mortality may be highly complex, and in patients hospitalized longer than 60 days, in spite of very high mortality, the ratio of deaths in patients admitted on an emergency basis to deaths in those admitted electively was only 1.23.

In summary, this was the first analysis of a large national database to assess the prognostic factors independently associated with in-hospital mortality in patients hospitalized in nonsurgical departments. The major strength of this study is the high number of analyzed hospitalizations (2855029), which makes the results more reliable than those obtained in smaller studies.

Limitations

However, our study had some limitations. First, we defined in-hospital mortality as death during hospitalization only. This was because the NFZ database does not contain information about deaths occurring after hospital discharge. Additionally, there is no consensus on the time after discharge during which death could still be related to the preceding hospital stay.⁴⁶ Therefore, since death during hospitalization is always recorded in hospital statistics, our approach to analyzing data on mortality seemed to be reliable and generated data that can be easily compared with those obtained in other studies.

Assessing only a few predictors of in-hospital nonsurgical mortality is another limitation, but we could analyze only factors available in the NFZ database. The choice of predictors was limited due to database structure and granularity of data. For example, hospital types were assigned to hospitalizations, yet not to the names of particular hospitals. This leads to an assumption that all hospitals of the same type have, on average, the same mortality rate. On one hand, this is a strong assumption, but, on the other hand, expecting that hospitals at the same reference level employ physicians providing services of the same quality sounds reasonable.

Analyzing in-hospital death in nonsurgical departments according to hospital types, we also could not address data clustering at the hospital level, which did not allow us to perceive the observation as fully independent (patients admitted to the same hospital could not be treated as independent of each other if more urgent admissions happened in particular hospitals).

Conclusions

In conclusion, age, male sex, emergency admission, admission on a particular day of the week or nonworking day (public holiday), and hospitalization in a district, city, private, or regional hospital (compared with a teaching hospital) are factors associated with greater mortality in nonsurgical departments.

Correspondence to

Prof. Edward Franek, MD, PhD, Department of Internal Diseases, Endocrinology and Diabetology, Central Clinical Hospital of the Ministry of the Interior and Administration in Warsaw, ul. Wołoska 137, 02-507 Warszawa, Poland, phone: +48225081405, email: edward.franek@cskmswia.pl

Received

November 23, 2019.

Revision accepted

February 9, 2020.

Published online

February 11, 2020.

Acknowledgments

This study was supported by the statutory funds of the Mossakowski Medical Research Centre.

Contribution statement

MW and EF conceived the concept of the study. All authors contributed to the research design and were involved in data collection. MCh analyzed the data. EF and MP-K coordinated funding of the project. MW and AT drafted the manuscript. All authors edited and approved the final version of the manuscript.

Conflict of interest

None declared.

How to cite

Walicka M, Chlebus M, Śliwczyński A, et al. Predictors of in-hospital mortality in nonsurgical departments: a multivariable regression analysis of 2855029 hospitalizations. Pol Arch Intern Med. 2020; 130: 268-275. doi:10.20452/pamw.15185

1.: Wright J, Dugdale B, Hammond I, et al. Learning from death: a hospital mortality reduction programme. J R Soc Med. 2006; 99: 303-308.Crossref
2.: Jarman B, Gault S, Alves B, et al. Explaining differences in English hospital death rates using routinely collected data. BMJ. 1999; 318: 1515-1520.Crossref
3.: .Ozierański K, Kapłon-Cieślicka A, Balsam P, et al. Effect of β-blockers on 1-year survival and hospitalizations in patients with heart failure and atrial fibrillation: results from ESC-HF Pilot and ESC-HF Long-Term Registry. Pol Arch Intern Med. 2018; 128: 649-657.
4.: .Kuźma Ł, Pogorzelski S, Struniawski K, et al. Effect of air pollution on the number of hospital admissions for acute coronary syndrome in elderly patients. Pol Arch Intern Med. 2020; 130: 38-46.
5.: Canadian Institute for Health Information. HSMR: A New Approach For Measuring Hospital Mortality Trends In Canada. Canadian Institute for Health Information; 2007. https://secure.cihi.ca/free_products/HSMR_hospital_mortality_trends_in_canada.pdf. Accessed August 28, 2017.
6.: Ben-Tovim D, Woodman R, Harrison J, et al. Measuring and reporting mortality in hospital patients. Australian Institute of Health and Welfare website. http://www.aihw.gov.au/publication-detail/?id=6442468221. Published March 13, 2009. Accessed January 16, 2016.
7.: Campbell MJ, Jacques RM, Fotheringham J, et al. Developing a summary hospital mortality index: retrospective analysis in English hospitals over five years. BMJ. 2012; 344: e1001.Crossref
8.: Vanhems P, Gambotti L, Fabry J. Excess rate of in-hospital death in Lyons, France, during the August 2003 heat wave. N Engl J Med. 2003; 349: 2077-2078.Crossref
9.: Campbell SE, Seymour DG, Primrose WR, et al. A multi-centre European study of factors affecting the discharge destination of older people admitted to hospital: analysis of in-hospital data from the ACMEplus project. Age Ageing. 2005; 34: 467-475.Crossref
10.: Campbell SE, Seymour DG, Primrose WR; ACMEPLUS Project. A systematic literature review of factors affecting outcome in older medical patients admitted to hospital. Age Ageing. 2004; 33: 110-115.Crossref
11.: Silva TJ, Jerussalmy CS, Farfel JM, et al. Predictors of in-hospital mortality among older patients. Clinics (Sao Paulo). 2009; 64: 613-618.Crossref
12.: Hwang LC, Hsu CP, Tjung JJ, et al. Predictors of in-hospital mortality in oldest-old patients in Taiwan. Int J Gerontol. 2013; 7: 22-26.Crossref
13.: Silva TJ, Sousa S, Moraes MF, et al. Predictive in-hospital and 6-month morbidity and mortality factors in elderly hospitalized patients. Acta Medica Portuguesa. 2002; 15: 177-184.
14.: Ponzetto M, Maero B, Maina P, et al. Risk factors for early and late mortality in hospitalized older patients: the continuing importance of functional status. J Gerontol A Biol Sci Med Sci. 2003; 58: 1049-1054.Crossref
15.: Mukhopadhyay A, Tai BC, See KC, et al. Risk factors for hospital and long-term mortality of critically ill elderly patients admitted to an intensive care unit. Biomed Res Int. 2014: 960575.Crossref
16.: Ayaz T, Sahin SB, Sahin OZ, et al. Factors affecting mortality in elderly patients hospitalized for nonmalignant reasons. J Aging Res. 2014: 584315.
17.: Haase KK, Schiele R, Wagner S, et al. In-hospital mortality of elderly patients with acute myocardial infarction: data from the MITRA (Maximal Individual Therapy in Acute Myocardial Infarction) registry. Clin Cardiol. 2000; 23: 831-836.Crossref
18.: Czach K, Klonowska K, Świderek M, Wiktorzak K. Poland: the Jednorodne Grupy Pacjentów – Polish experiences with DRGs. In: Busse R, Geissler A, Quentin W, Wiley M, eds. Diagnosis-Related Groups in Europe. Moving towards Transparency, Efficiency and Quality in Hospitals. Maidenhead, United Kingdom: Open University Press; 2011: 259-379.
19.: Allison PD. Logistic Regression Using the SAS System: Theory and Application. Cary, North Carolina, United States: SAS Institute; 1999.
20.: Ani C, Pan D, Martins D, Ovbiagele B. Age- and sex-specific in-hospital mortality after myocardial infarction in routine clinical practice. Cardiol Res Pract. 2010: 752765.Crossref
21.: Nowakowska-Arendt A, Grąbczewska Z, Koziński M, et al. Gender differences and in-hospital mortality in patients undergoing percutaneous coronary interventions. Kardiol Pol. 2008; 66: 632-639.
22.: Milcent C, Dormont B, Durand-Zaleski I, Steg PG. Gender differences in hospital mortality and use of percutaneous coronary intervention in acute myocardial infarction. Microsimulation analysis of the 1999 Nationwide French Hospitals Database. Circulation. 2007; 115: 833-839.Crossref
23.: Zhang Z, Fang J, Gillespie C, et al. Age-specific gender differences in in-hospital mortality by type of acute myocardial infarction. Am J Cardiol. 2012; 109: 1097-1103.Crossref
24.: Numasawa Y, Inohara T, Ishii H, et al. Comparison of outcomes of women versus men with non-ST-elevation acute coronary syndromes undergoing percutaneous coronary intervention (from the Japanese Nationwide Registry). Am J Cardiol. 2017; 119: 826-831.Crossref
25.: Sweileh WM, Sawalha AF, Al-Aqad SM, et al. Predictors of in-hospital mortality after acute stroke: impact of gender. Int J Clin Exp Med. 2009; 2: 41-47.
26.: Gulácsi L, Májer I, Kárpáti K, et al. Mortality of hospitalized stroke patients in Hungary; 2003-2005. Ideggyogy Sz. 2007; 60: 321-328.
27.: Gordon HS, Rosenthal GE. The relationship of gender and in-hospital death: increased risk of death in men. Med Care. 1999; 37: 318-324.Crossref
28.: Staneva M. In-hospital mortality in Mississippi, 2010. Mississippi State Department of Health website. http://msdh.ms.gov/msdhsite/index.cfm/31,5512,348,pdf/MS_Hospital_Mortality2010.pdf. Published 2013. Accessed August 25, 2017.
29.: Cereda E, Klersy C, Hiesmayr M, et al. NutritionDay survey collaborators. Body mass index, age and in-hospital mortality: the NutritionDay multinational survey. Clin Nutr. 2017; 36: 839-847.Crossref
30.: Deaths in Australian hospitals 2014–15. Australian Institute of Health and Welfare website. https://www.aihw.gov.au/reports/hospitals/deaths-in-australian-hospitals-2014-15/contents/summary. Accessed August 28, 2017.
31.: Zelada H, Bernabe-Ortiz A, Manrique H. In-hospital mortality in patients with type 2 diabetes mellitus: a prospective cohort study in Lima, Peru. J Diabetes Res. 2016: 7287215.Crossref
32.: Nayar P, Yu F, Chandak A, et al. Risk factors for in-hospital mortality in heart failure patients: does rurality, payer or admission source matter? J Rural Health. 2018; 34: 103-108.Crossref
33.: Cram P, Hillis SL, Barnett M, Rosenthal GE. Effects of weekend admission and hospital teaching status on in-hospital mortality. Am J Med. 2004; 117: 151-157.Crossref
34.: Burke LG, Frakt AB, Khullar D, et al. Association between teaching status and mortality in US hospitals. JAMA. 2017; 317: 2105-2113.Crossref
35.: Cid Pedraza C, Herrera CA, Prieto Toledo L, Oyarzún F. Mortality outcomes in hospitals with public, private not-for-profit and private for-profit ownership in Chile 2001-2010. Health Policy Plan. 2015; 30 (suppl 1): i75-i81.Crossref
36.: Suissa S, Dell’Aniello S, Suissa D, Ernst P. Friday and weekend hospital stays: effects on mortality. Eur Respir J. 2014; 44: 627-633.Crossref
37.: Gallerani M, Fedeli U, Pala M, et al. Weekend versus weekday admission and in-hospital mortality for pulmonary embolism: a 14-year retrospective study on the national hospital database of Italy. Angiology. 2018; 69: 236-241.
38.: Walker AS, Mason A, Quan TP, et al. Mortality risks associated with emergency admissions during weekends and public holidays: an analysis of electronic health records. Lancet. 2017; 390: 62-72.
39.: Hoshijima H, Takeuchi R, Mihara T, et al. Weekend versus weekday admission and short-term mortality: a meta-analysis of 88 cohort studies including 56,934,649 participants. Medicine (Baltimore). 2017; 96: e6685.
40.: Conway R, Cournane S, Byrne D, et al. Improved mortality outcomes over time for weekend emergency medical admissions. Ir J Med Sci. 2018; 187: 5-11.
41.: Sayers A, Whitehouse MR, Berstock JR, et al. The association between the day of the week of milestones in the care pathway of patients with hip fracture and 30-day mortality: findings from a prospective national registry - The National Hip Fracture Database of England and Wales. BMC Med. 2017; 15: 62.
42.: Isogai T, Matsui H, Tanaka H, et al. Seasonal variation in patient characteristics and in-hospital outcomes of: a nationwide retrospective cohort study in Japan. Heart Vessels. 2017; 32: 1271-1276.
43.: Siddiqui T, Mehta D, Haq KF, et al. Seasonal effect on incidence and in-hospital mortality of acute bowel ischemia. Am J Gastroenterol. 2018; 113: S84-S85.
44.: Vicent L, Bruña V, Devesa C, et al. Seasonality in mortality in a cardiology department: a five-year analysis in 500 patients. Cardiology. 2019; 142: 67-72.
45.: Hall MJ, Levant S, DeFrances CJ. Trends in inpatient hospital deaths: National Hospital Discharge Survey, 2000-2010. NCHS Data Brief. 2013; 118: 1-8.
46.: Brinkman S, de Jonge E, Abu-Hanna A, et al. Mortality after hospital discharge in ICU patients. Crit Care Med. 2013; 41: 1229-36.