Factors associated with pressure ulcers among middle-old and oldest-old inpatients in an acute care hospital

Abstract

With the aging population, the majority of patients in acute care hospitals in high-income countries are older adults. To identify the factors associated with pressure ulcers among patients aged ≥ 75 years in an acute care hospital,we conducted a retrospective study using individual-level secondary data obtained from patient medical and nursing records. The study population was patients aged ≥ 75 years who were admitted to the hospital between March 2016 and February 2021. We fitted multivariable logistic regression models. A total of 15,258 were included in the analysis, and 513 (3.4%) patients were with pressure ulcers. The factors associated with pressure ulcers were Body Mass Index (adjusted OR = 0.96, 95% CI: 0.93–0.98); albumin level (0.51, 0.42–0.63); cardiovascular disease (2.10, 1.57–2.82); infection (1.91, 1.07–3.42); bedridden (1.85, 1.01–3.39); using bed when moving (0.45, 0.25–0.84); fecal or urine incontinence (2.44, 1.46–4.06); admission to the emergency medicine (1.71, 1.06–2.74); length of stay (1.02, 1.01–1.02); and tube feeding (2.02, 1.44–2.82). Low nutritional status, low activity levels, incontinence, cardiovascular disease, infection, admission to the emergency medicine, and more extended hospital stay were significantly associated with pressure ulcers. Therefore, nurses in acute care should carefully monitor the patients aged ≥ 75 years with these factors.

Introduction

With the aging population, the majority of patients in acute care hospitals in high-income countries are older adults. For instance, in US and Canada, older adults (aged ≥ 65 years) account for approximately 30% of all patients receiving a surgical intervention and nearly half of all acute inpatients¹. In Japan, adults aged 65 years and older and those aged 75 years and older accounted for 73.3% and 53.1% of all hospital admissions, and the proportion of older adults of all acute inpatients is also on the rise². When older adults are hospitalized for acute care, they are at a higher risk for postoperative complications, readmission, delirium, and pressure ulcers^3–8.

In particular, the risk of pressure ulcers (also known as pressure injuries, bedsores, pressure sores, or decubitus ulcers) is increased among older inpatients^9–14. Furthermore, pressure ulcers can cause more harm than the presenting clinical conditions^15–18. Pressure ulcers are sections of tissue that sustain irreversible ischemic damage from an external force that reduces or blocks blood flow to the soft tissues between the bones and the surface layer of the skin for a certain period of time¹⁹. Pressure ulcer development is associated with multiple factors, including age, patient activity and mobility, skin moisture, and perceived cognition²⁰. Age is one of the important associated factors with pressure ulcers, suggesting that the risk of pressure ulcers increases with older age^9,10. However, little is known about the factors for pressure ulcers among inpatients of the old and oldest-old (those aged ≥ 75 years) in acute care, which has increased rapidly in recent years^21,22.

Since the risk of pressure ulcers is a potential determinant of care resource use, it should inform service provision planning and resource allocation; however, there are major barriers to the identification of this risk among older patients admitted to acute care hospitals. While there are tools available to measure the risk of pressure ulcers^20,23–26, most tools are difficult to use and time-consuming to assess for pressure ulcers in acute care hospitals^27–29. Therefore, there is a need for an automated system for identifying older patients at high risk for pressure ulcers using existing medical record data^27,30. However, a dearth of evidence remains regarding what data from regular inpatient medical records are associated with pressure ulcers.

Among older adults aged ≥ 75 years in chronic care hospitals, nursing homes, and home care, previous studiers shows that incontinence, bedridden, or receiving tube feeding were more likely to have pressure ulcers^31–35. However, little is known about the factors for pressure ulcers among inpatients aged ≥ 75 years in acute care. Therefore, we aimed to identify factors associated with pressure ulcers among patients of the old and oldest-old (those aged ≥ 75 years) in an acute care hospital using data from regular inpatient medical records.

Methods

Study design

This retrospective study was designed using individual-level secondary data obtained from patient medical and nursing records provided by the University of Osaka Hospital in Japan. This university hospital currently staffs 1,145 nurses and is committed to providing quality healthcare and training outstanding nurses. In 2019, the number of inpatients was 336,410, with an average hospital stay of 14.4 days³⁶.

Participants

The study population was patients who were admitted to the university hospital between March 2016 and February 2021. The criteria for inclusion into the study were: (1) age 75 years or older, (2) admission to the university hospital, and (3) hospitalized between March 2016 and February 2021. Patients with missing data were excluded from the analysis.

Data collection

Data were collected using electronic medical records of the University of Osaka Hospital in Japan. We obtained individual-level secondary data between March 2016 and February 2021. We identified potentially relevant variables as previous studies, and the following variables were extracted from the electronic medical records: age^{9,10,14,37–40}; sex^{14,28,39,41–43}; body mass index [BMI]^44,45; primary diagnosis (i.e., cancer, cardiovascular disease, diabetes mellitus, infection, other)^{10,31,38,42,46}; Japan Coma Scale (i.e., level 0 [no problem], level 1 [delirium, confusion, senselessness], level 2 [stupor, lethargy, hypersomnia, somnolence, drowsiness], level 3 [deep coma, coma, semi-coma])^28,48; albumin^40,41,49; total protein⁴⁴; hemoglobin^37,49; activity level (i.e., unrestricted activity, walking to toilets, sitting up in bed, bedridden)⁵⁰, transferring (i.e., walking, cane or rollator, wheelchair, bed)^28,40; toileting (i.e., toilet use, use of portable toilet, fecal or urine incontinence)^39,49; and history of smoking (yes or no)^51,52. As for the hospital environment and treatment, the following variables were extracted: type of admission (unscheduled admission or scheduled admission)⁵³; hospital ward (i.e., surgery, internal medicine, sensory, skin, and motor function, emergency medicine, other)^28,39; length of stay [LOS]^{39,40,42,46,49,50,54,55}; surgical treatment during stay (yes or no)¹⁴; bed rest as a treatment during stay (yes vs. no)^14,56,57; use of feeding tube during stay (yes or no)^38,39; parenteral nutrition during stay (yes or no)^38,39; and respirator use during stay (yes or no)^32,43.

Data analysis

Using ‘pressure ulcer’ (bivariate) as the dependent variable, we fitted multivariable logistic regression models to determine the factors associated with pressure ulcers among patients of the old and oldest-old (those aged ≥ 75 years) in an acute care hospital. All odds ratios (ORs) were calculated with regard to the identified potentially relevant variables: basic patient information, primary diseases, coma levels, nutritional status, activity levels, transferring, toileting, smoking, types of admission, hospital wards, length of stay, surgical treatment, bed rest as a treatment, use of feeding tube, parental nutrition, and respirator use. All potentially relevant variables were included in the models unless there was strong evidence of collinearity, as reflected by a variance inflation factor greater than 10. List-wise deletion was used to address missing values. All reported model p values were considered significant at p < 0.05. All analyses were performed using Stata 16 (StataCorp. 2019. Stata Statistical Software: Release 16. College Station, TX: StataCorp LLC, https://www.stata.com/stata16/).

Validity and reliability

In the sensitivity analysis, we restricted the patients of the LOS ≤ 3 days, excluding short-term hospitalizations that are less likely to be related to pressure ulcers⁴⁹. Moreover, we restricted the patients with BMI > 30, excluding obese, because BMI and the likelihood of having pressure ulcers have a U-shaped relationship^39,58.

Ethical consideration

Ethical considerations were in accordance with the Japanese epidemiological guidelines for secondary data analysis. Ethical approval for the study was obtained from the institutional review board at the University of Osaka Hospital in Japan (approval number: 20569). The study used de-identified data (fully anonymized prior to receipt of the data); therefore, the need for obtaining written informed consent was waived by the Institutional Review Board of the University of Osaka Hospital, Japan.

Results

Patient characteristics

Of the 23,210 patients identified, 15,258 met the inclusion criteria and were included in the analysis. Total blood protein was the most frequently missing variable, absent in 3,352 patients (14.4%), followed by albumin, which was missing in 2,581 patients (11.1%). Among the 15,258 patients analyzed, 7,989 (52.4%) were male. The median age was 79 years (interquartile range [IQR]: 77–83), and the mean body mass index (BMI) was 21.9 (standard deviation [SD]: 3.5). A total of 4,609 patients (30.2%) had cancer, and 3,563 (23.4%) had cardiovascular disease (Table 1). Regarding the hospital environment, 597 (2.3%) were admitted to the ward of emergency medicine, and the median LOS was 12 (6–22) days. The proportions of patients with and without pressure ulcers were 3.4% and 96.6%, respectively.

Table 1.

Characteristics of patients with and without pressure ulcers (n = 15,258).

	With pressure ulcer (n = 513)		Without pressure ulcer (n = 14,745)		Total (n = 15,258)
	n	(%)	n	(%)	n	(%)
Patient characteristics
Age (years), median (IQR)	80	(77–84)	79	(77–83)	79	(77–83)
Sex, n (%)
Female	234	47.7	7035	45.6	7269	47.6
Male	279	52.3	7710	54.4	7989	52.4
Body mass index (kg/m2), mean (SD)	20.7	3.6	21.9	3.5	21.9	3.5
Primary disease, n (%)
Cancer	106	20.7	4503	30.5	4609	30.2
Cardiovascular disease	190	37.0	3373	22.9	3563	23.4
Diabetes mellitus	4	0.8	321	2.2	325	2.1
Infection	27	5.3	201	1.4	228	1.5
Other	186	36.3	6347	43.1	6533	42.8
Japan Coma Scale, n (%)
Level 0	406	79.1	14,228	96.5	14,634	95.9
Level 1	54	10.5	285	1.9	339	2.2
Level 2	13	2.5	82	0.6	95	0.6
Level 3	40	7.8	150	1.0	190	1.3
Nutrition (g/dL), mean (SD)
Albumin	3.3	0.7	3.8	0.5	3.8	0.5
Total protein	6.5	1.0	6.9	0.7	6.9	0.7
Hemoglobin	11.4	2.1	12.2	1.9	12.1	1.9
Activity, n (%)
Unrestricted activity	78	15.2	5254	35.6	5332	35.0
Walking to toilets	189	36.8	6092	41.3	6281	41.2
Sitting up in bed	27	5.3	583	4.0	610	4.0
Bedridden	219	42.7	2816	19.1	3035	19.9
Transferring, n (%)
Walking	107	20.9	6208	42.1	6315	41.4
Cane or rollator	5	1.0	222	1.5	227	1.5
Wheelchair	179	34.9	5272	35.8	5451	35.7
Bed	222	43.3	3043	20.6	3265	21.4
Toileting, n (%)
Toilet use	264	51.5	11,496	78.0	11,760	77.1
Use of portable toilet	8	1.6	136	0.9	144	0.9
Fecal or urine incontinence	241	47.0	3113	21.1	3354	22.0
Smoking, n (%)
Yes	134	26.1	3946	26.8	4080	26.7
No	379	73.9	10,799	73.2	11,178	73.3
Hospital environment and treatment
Admission type, n (%)
Unscheduled admission	205	40.0	2070	14.0	2275	14.9
Scheduled admission	308	60.0	12,675	86.0	12,983	85.1
Hospital ward, n (%)
Surgery	136	26.5	3271	22.2	3407	29.7
Internal medicine	135	26.3	4918	33.4	5053	38.6
Sensory and motor function	61	11.9	3915	26.6	3976	26.9
Emergency medicine	104	20.3	493	3.3	597	2.3
Other	77	15.0	2148	14.6	2225	2.4
Length of stay (days), median (IQR)	23	(11–42)	12	(6–21)	12	(6–22)
Surgical treatment during stay, n (%)
Yes	198	38.6	6365	43.2	6563	43.0
No	315	61.4	8380	56.8	8695	57.0
Bed rest as a treatment during stay, n (%)
Yes	349	68.0	7283	49.4	7632	50.0
No	164	32.0	7462	50.6	7626	50.0
Use of feeding tube during stay, n (%)
Yes	124	24.2	446	3.0	570	3.7
No	389	75.8	14,299	97.0	14,688	96.3
Parenteral nutrition during stay, n (%)
Yes	84	16.4	741	5.0	825	5.4
No	429	83.6	14,004	95.0	14,433	94.6
Respirator use during stay, n (%)
Yes	153	29.8	1026	7.0	1179	7.7
No	360	70.2	13,719	93.0	14,079	92.3

IQR, interquartile range; SD, standard deviation.

Among the 513 (3.4%) patients with pressure ulcers, the mean BMI was 20.7 (3.6), 190 (37.0%) had cardiovascular diseases, 106 (20.7%) had cancers, 27 (5.3%) had infections, 107 (20.8%) were unresponsiveness or coma, the mean albumin was 3.3 (0.7), 219 (42.7%) were bedridden, and 241 (27.0%) were fecal or urine incontinence. In addition, 104 (20.3%) were admitted to the ward of emergency medicine, the median LOS was 23 (11–42) days, and 153 (29.8%) required respirators.

Among the 14745 (96.6%) patients without pressure ulcers, the mean BMI was 21.9 (3.5), 4503 (30.5%) had cancers, 3373 (22.9%) had cardiovascular diseases, 201 (1.4%) had infections, 517 (3.5%) were unresponsiveness or coma, and the mean albumin was 3.8 (0.5), 2816 (19.1%) were bedridden, and 3113 (21.1%) were fecal or urine incontinence. Furthermore, 104 (20.3%) were admitted to the ward of emergency medicine, the median LOS was 23 (11–42) days, and 153 (29.8%) required respirators.

Logistic regression analysis

Regarding the multilevel logistic regression analysis, the factors associated with pressure ulcers were BMI (adjusted OR = 0.96, 95% CI: 0.93–0.98); cardiovascular disease (2.10, 1.57–2.82); infection (1.91, 1.07–3.42); Japan Come Scale level 1 (1.58, 1.01–2.48); albumin level (0.51, 0.42–0.63); walking to toilets (1.52, 1.02–2.27), bedridden (1.85, 1.01–3.39); using bed when moving (0.45, 0.25–0.84); fecal or urine incontinence (2.44, 1.46–4.06); admission to the emergency medicine (1.71, 1.06–2.74); LOS (1.02, 1.01–1.02); and tube feeding (2.02, 1.44–2.82). A summary of all factors considered and their corresponding p values in the multivariable logistic regression are shown in Table 2. We found that low BMI, cardiovascular disease, infection, Japan Come Scale level 1, a low albumin level, walking to the toilet, being bedridden, using a bed when moving, fecal or urinary incontinence, emergency ward admission, LOS, and tube feeding were significantly associated with pressure ulcers.

Table 2.

Logistic regression analysis of the factors associated with pressure ulcers (n = 15,258).

	Univariable		Multivariable
	Adjusted OR (95% CI)	p	Adjusted OR (95% CI)	p
Patient characteristics
Age	1.04 (1.02–1.06)	< 0.01	1.01 (0.99–1.04)	0.20
Sex
Female	1 (ref)		1 (ref)
Male	1.09 (0.91–1.30)	0.35	1.05 (0.85–1.29)	0.67
Body mass index	0.90 (0.88–0.93)	< 0.01	0.96 (0.93–0.98)	< 0.01
Primary disease
Cancer	1 (ref)		1 (ref)
Cardiovascular disease	2.39 (1.88–3.05)	< 0.01	2.10 (1.57–2.82)	< 0.01
Diabetes mellitus	0.53 (0.19–1.45)	0.22	0.75 (0.27–2.09)	0.58
Infection	5.71 (3.65–8.91)	< 0.01	1.91 (1.07–3.42)	0.03
Other	1.24 (0.98–1.59)	0.08	1.28 (0.96–1.71)	0.09
Japan Coma Scale
Level 0	1 (ref)		1 (ref)
Level 1	6.64 (4.88–9.03)	< 0.01	1.58 (1.01–2.48)	0.04
Level 2	5.56 (3.07–10.06)	< 0.01	1.26 (0.63–2.50)	0.51
Level 3	9.35 (6.50–13.43)	< 0.01	1.22 (0.71–2.09)	0.48
Nutrition status
Albumin	0.31 (0.28–0.36)	< 0.01	0.51 (0.42–0.63)	< 0.01
Total protein	0.51 (0.46–0.57)	< 0.01	0.94 (0.82–1.08)	0.37
Hemoglobin	0.83 (0.80–0.87)	< 0.01	1.01 (0.96–1.07)	0.66
Activity
Unrestricted activity	1 (ref)		1 (ref)
Walking to toilets	2.09 (1.60–2.73)	< 0.01	1.52 (1.02–2.27)	0.04
Sitting up in bed	3.12 (2.00–4.87)	< 0.01	1.40 (0.75–2.63)	0.29
Bedridden	5.24 (4.03–6.81)	< 0.01	1.85 (1.01–3.39)	0.05
Transferring
Walking	1 (ref)		1 (ref)
Cane or rollator	1.31 (0.53–3.24)	0.56	0.79 (0.30–2.08)	0.64
Wheelchair	1.97 (1.55–2.51)	< 0.01	1.02 (0.70–1.48)	0.92
Bed	4.23 (3.35–5.35)	< 0.01	0.45 (0.25–0.84)	0.01
Toileting
Toilet use	1 (ref)		1 (ref)
Use of portable toilet	2.56 (1.24–5.28)	0.01	1.26 (0.59–2.71)	0.55
Fecal or urine incontinence	3.37 (2.82–4.03)	< 0.01	2.44 (1.46–4.06)	< 0.01
Smoking (yes vs. no)	0.97 (0.79–1.18)	0.75	0.92 (0.72–1.18)	0.52
Hospital environment and treatment
Unscheduled admission (yes vs. no)	4.08 (3.40–4.89)	< 0.01	0.91 (0.67–1.25)	0.57
Hospital ward
Surgery	1 (ref)		1 (ref)
Internal medicine	0.66 (0.52–0.84)	< 0.01	0.91 (0.67–1.23)	0.55
Sensory and motor function	0.37 (0.28–0.51)	< 0.01	0.87 (0.59–1.29)	0.50
Emergency medicine	5.07 (3.86–6.66)	< 0.01	1.71 (1.06–2.74)	0.03
Other	0.86 (0.65–1.15)	0.31	1.22 (0.86–1.73)	0.27
Length of stay	1.03 (1.02–1.03)	< 0.01	1.02 (1.01–1.02)	< 0.01
Surgical treatment (yes vs. no)	0.83 (0.69–0.99)	0.04	0.88 (0.66–1.16)	0.36
Bed rest as a treatment (yes vs. no)	2.18 (1.81–2.63)	< 0.01	1.20 (0.93–1.54)	0.16
Use of feeding tube (yes vs. no)	10.22 (8.18–12.77)	< 0.01	2.02 (1.44–2.82)	< 0.01
Parenteral nutrition (yes vs. no)	3.70 (2.90–4.73)	< 0.01	1.36 (0.93–1.98)	0.12
Respirator use (yes vs. no)	5.68 (4.66–6.94)	< 0.01	1.06 (0.75–1.50)	0.74

The multivariable ORs were adjusted for all variables.

CI, confidence interval; OR, odds ratio.

Results of sensitivity analyses

First, we restricted patients of LOS ≤ 3 days, excluding 993 (6.5%) patients for the multilevel logistic regression analysis. The sensitivity analysis among patients of LOS ≤3 days showed that the factors significantly associated with pressure ulcers were BMI (0.95, 0.93–0.98); cardiovascular disease (2.17, 1.60–2.92); infection (2.11, 1.16–3.82); albumin level (0.52, 0.42–0.64); walking to toilets (1.61, 1.07–2.41); bedridden (2.15, 1.13–4.07); use a bed when moving (0.44, 0.24–0.83); fecal or urine incontinence (2.31, 1.37–3.92); admission to the emergency medicine (1.71, 1.05–2.79); LOS (1.02, 1.01–1.02); and tube feeding (1.90, 1.34–2.68).

Second, we restricted patients with BMI > 30, excluding 258 (1.7%) patients for the multilevel logistic regression analysis. The sensitivity analysis among patients with BMI > 30 showed that the factors significantly associated with pressure ulcers were BMI (0.96, 0.93–0.99); cardiovascular disease (2.12, 1.58–2.84); infection (1.87, 1.04–3.37); albumin level (0.51, 0.42–0.63); walking to toilets (1.54, 1.03–2.31); bedridden (1.90, 1.03–3.51); use a bed when moving (0.44, 0.24–0.81); fecal or urine incontinence (2.47, 1.48–4.13); admission to the emergency medicine (1.72, 1.07–2.78); LOS (1.02, 1.01–1.02); and tube feeding (2.01, 1.43–2.81).

Discussion

We conducted multivariable logistic regression analyses to identify the factors associated with pressure ulcers among patients aged ≥ 75 years in an acute care hospital using data from regular inpatient medical records. We found that low BMI, cardiovascular disease, infection, Japan Come Scale level 1, a low albumin level, walking to toilets, bedridden, using bed when moving, fecal or urine incontinence, admission to the emergency medicine, LOS, and tube feeding were significantly associated with pressure ulcers.

We found that patients aged ≥ 75 years in acute care who were admitted to the ward of emergency medicine, with a longer hospital stay, cardiovascular diseases, or infections were more likely to have pressure ulcers as well as among all adult patients receiving acute care^{46,48–50,59}. Notably, the odds of having pressure ulcers were nearly twice as high among patients with cardiovascular diseases or infections compared to those with cancer. These findings were consistent with prior studies reporting an increased risk of pressure ulcers in patients with cardiovascular disease^60,61. Specifically, individuals with left main or triple-vessel coronary lesions exhibit higher susceptibility to pressure ulcers⁶⁰. Patients undergoing cardiac surgery were also considered had high-risk, with contributing factors such as prolonged operative time, intraoperative hypothermia, vascular pathology, and postoperative use of vasoconstrictive agents, all of which may impair skin and soft tissue perfusion, increasing pressure ulcer risk⁶¹. These results underscore the need for careful pressure ulcer monitoring in acute care settings for patients with cardiovascular diseases, infections, prolonged hospitalization, or emergency admissions, regardless of age.

Patients aged ≥ 75 years in acute care who were incontinence, bedridden, or receiving tube feeding were more likely to have pressure ulcers, which are in line with previous evidence in chronic care hospitals, nursing homes, and home care^31–35. In this study, the odds of having pressure ulcers was 2.44 in incontinence as without incontinence. Incontinence, bedridden, and tube feeding are important factors associated with pressure ulcers among older adults aged ≥ 75 years, regardless of where they receive care. These findings indicate that healthcare providers and caregivers of older adults aged ≥ 75 years should pay sufficient attention to pressure ulcers when the older adults are incontinence, bedridden, or receiving tube feeding.

Patients who had low BMI or a low albumin level were more likely to have pressure ulcers among patients aged ≥ 75 years in acute care. In particular, the odds of having a pressure ulcer decreased by about half when albumin levels increased by one. These findings are consistent with those of previous studies that showed that patients with malnutrition and underweight were more likely to develop pressure ulcers^41,43, which suggests that assessments of albumin level and BMI are essential to identify patients with pressure ulcers even in the patients aged ≥ 75 years. This study has limitations as well as strengths. First, owing to the observational study design and analysis of secondary data, no data on several potential confounders, including skin status, were obtained^20,62,63. However, the aim of this study was to identify factors associated with pressure ulcers using data from existing inpatient electronic medical records. The data obtained from inpatient electronic medical records were beneficial in elucidating the factors associated with pressure ulcers in older inpatients. Second, because this study was conducted on patients admitted to a single university hospital, it is uncertain how our findings may apply to patients in other acute care hospitals and other countries. Further studies should be conducted on patients in a wide variety of acute care hospitals and examine the generalizability of these findings. Third, acute care hospitals in Japan have a nurse dedicated to the management of pressure ulcers, who assists nurses in the wards with pressure ulcer care. Moreover, all patients were assessed for pressure ulcers and received preventive measures, which may have influenced the results. However, to the best of our knowledge, this is the first study to investigate the factors associated with pressure ulcers among inpatients aged ≥ 75 years in an acute care hospital using data from inpatient electronic medical records.

Conclusions

We examined the factors associated with pressure ulcers among patients aged ≥ 75 years in an acute care hospital using data from regular inpatient medical records. Low BMI, cardiovascular disease, infection, Japan Come Scale level 1, a low albumin level, walking to toilets, bedridden, using bed when moving, fecal or urine incontinence, admission to the emergency medicine, length of hospital stay, and tube feeding were significantly associated with pressure ulcers among inpatients aged ≥ 75 years in acute care. Therefore, patients aged ≥ 75 year with these factors associated with pressure ulcers should be carefully monitored by nurses in acute care.

Abbreviations

BMI

Body mass index

CI

Confidence interval

IQR

Interquartile range

LOS

Length of stay

OR

Odds ratio

SD

Standard deviation

Author contributions

All authors contributed substantially to the conception and design; ST, KA, KK, RN, YN, TM, SM and TT collected data; SI drafted the article; SI performed the statistical analysis; SI, SY, RY, ST, KA, KK, TT and YM interpreted the data; all authors critically revised the article for important intellectual content; and all authors made meaningful contributions to the final submitted manuscript.

Data availability

The data that support the findings of this study are not openly available due to reasons of sensitivity and are available from the corresponding author upon reasonable request and with permission from the University of Osaka Hospital in Japan.

Declarations

Ethics approval and consent to participate

Ethical considerations were in accordance with the Japanese epidemiological guidelines for secondary data analysis. Ethical approval for the study was obtained from the institutional review board at Osaka University Hospital in Japan (approval number: 20569). The study used de-identified data (fully anonymized prior to receipt of the data); therefore, participant consent was exempted by the Institutional Review Board of Osaka University Hospital, Japan.

Competing interests

SI, KK, and MK: Department of Genome Informatics is an endowment department, supported with an unrestricted grant from NEC Corporation. All other authors declare no competing interests.