Requirements for hospitals in Japan to have low operative mortality and failure‐to‐rescue rates

Abstract

Aim

We explored institutional factors in Japan associated with lower operative mortality and failure‐to‐rescue (FTR) rates for eight major gastrointestinal procedures.

Methods

A 22‐item online questionnaire was sent to 2119 institutional departments (IDs) to examine the association between institutional factors and operative mortality and FTR rates. IDs were classified according to the number of annual surgeries, board certification status, and locality. In addition, the top 20% and bottom 20% of IDs were identified based on FTR rates and matched with the results of the questionnaire survey. Factors associated with operative mortality were selected by multivariate analysis.

Results

Of the 1083 IDs that responded to the questionnaire, 568 (213 382 patients) were included in the analysis. Operative morbidity, operative mortality, and FTR rates in the top 20% and bottom 20% of IDs were 13.1% and 8.4% (p < 0.001), 0.52% and 4.3% (p < 0.001), and 4.0% and 51.2% (p < 0.001), respectively. Based on the patients' background characteristics, the top 20% of IDs handled more advanced cases. No significant difference in locality was seen between better or worse hospital FTR rates, but fewer esophagectomies, hepatectomies, and pancreatoduodenectomies were performed in depopulated areas. Six items were found to be associated with operative mortality by multivariate logistic analysis. Only 50 (8.8%) IDs met all five factors related to better FTR rates.

Conclusions

The present findings indicate that several hospital factors surrounding surgical treatment, characterized by abundant human resources, are closely related to better postoperative recovery from severe complications.

We explored institutional factors in Japan associated with lower operative mortality and failure‐to‐rescue (FTR) rates for eight major gastrointestinal procedures. Operative morbidity, operative mortality, and FTR rates in the top 20% and bottom 20% of IDs were 13.1% and 8.4% (p < 0.001), 0.52% and 4.3% (p < 0.001), and 4.0% and 51.2% (p < 0.001), respectively. Several hospital factors surrounding surgical treatment, characterized by abundant human resources, are closely related to better postoperative recovery from severe complications.

1. INTRODUCTION

Because of its invasive nature, surgery is associated with injury to the human body and may result in a certain frequency of postoperative complications. ¹ Furthermore, in patients who do not recover from morbidity, the worsening of their condition may eventually lead to death. Although many efforts have been made to reduce operative mortality, including the identification of risk factors, improvement of surgical techniques, and better communication among medical professionals, zero mortality has only been achieved in a very limited number of hospitals.

There have been various reports on hospital factors that influence operative mortality. Operative mortality and the annual number of operations are known to have an inverse correlation (hospital volume). ² , ³ Furthermore, the degree of correlation varies greatly among surgical procedures. ⁴ , ⁵ Numerous studies have attempted to elucidate the factors associated with improved surgical safety in high‐volume centers. For example, Konno et al. ⁶ reported that operative mortality decreased as the number of the following 10 institutional evaluation items were met: “elective surgery is decided in a preoperative conference,” “a cancer board is held,” “a mortality and morbidity conference is held,” “a National Clinical Database (NCD) feedback system is used for clinical treatment,” “a team treatment system is built,” “information and communications technology is installed,” “a nutrition support team is installed,” “the World Health Organization surgical safety checklist is checked when starting surgery,” “there are two or more board‐certified surgeons,” and “there is a certified nurse.” However, why hospital volume has a significant impact remains unclear.

Silber et al. ⁷ reported that mortality rates were associated with both hospital and patient characteristics. On the other hand, failure‐to‐rescue (FTR) rates were more strongly associated with hospital characteristics and less influenced by the patient background. Ghaferi et al. ⁸ reported that the best (low mortality) and worst (high mortality) hospitals had similar postoperative morbidity and FTR rates. A systematic review also reported that FTR rates were inversely related to hospital volume. ⁹ Ward et al. ¹⁰ identified several institutional factors associated with FTR rates, including the proportion of board‐certified intensivists, whether a closed intensive care unit (ICU) model was in place, the presence of advanced practice providers such as nurse practitioners and physician assistants, and the presence of rapid response teams.

Lower mortality is clearly associated with greater attention to the timely recognition and multidisciplinary management of complications once they occur. If such factors could be clearly identified, healthcare providers could adopt this information to achieve better operative outcomes. Given this background, the present study aimed to clarify the institutional and human resources associated with better or worse operative mortality and FTR rates using data from the NCD in Japan.

2. MATERIALS AND METHODS

This study was approved by the institutional review board of Yokohama City University (Approval nos. B190400030 and B200806086).

2.1. National Clinical Database

Patient data collection via the Internet began in 2011, and annually, data from more than 1.4 million patients are collected from approximately 4000 hospitals in Japan. The NCD covers more than 95% of all surgeries in Japan. The present study used data from all patients who underwent eight representative procedures between January 1, 2016, and December 31, 2018. The main outcome measure of this study was operative mortality, defined as death within the index hospitalization period, regardless of the length of hospital stay (trackable up to 90 days), and the FTR rate, defined as death in a patient with at least one postoperative complication requiring an invasive intervention (Clavien–Dindo classification Grade III or higher). ¹¹ We determined the FTR rate for hospitals by evaluating the proportion of deaths among patients who developed a postoperative complication (numerator) to the total number of patients who developed a postoperative complication (denominator). Subsequently, the top 20% of IDs with a good FTR rate (i.e., the best hospitals) were compared with the bottom 20% of IDs with a poor FTR rate (i.e., the worst hospitals).

2.2. Board certification system of the JSGS

The board certification system of the JSGS is composed of approximately 8600 board‐certified gastrointestinal surgeons and 976 board‐certified training institutions.

2.3. Locality of the hospitals

Japan is divided into 47 administrative prefectures. The prefectures are further composed of secondary medical care regions, which are areas set up to complete general inpatient treatment, including emergency care. As of September 2020, Japan had 335 secondary medical areas. Based on a survey by Takahashi, ¹² , ¹³ three regional classifications have been defined according to population density. Briefly, secondary medical areas with “a population of ≥ 1 million or a population density of 2000 people per square kilometer” are classified as “Metropolitan areas,” whereas those with “a population of ≥ 200 000 or a population of 100 000–200 000 with a population density of ≥200 people per square kilometer” are classified as “Provincial city areas.” Areas that do not meet either of these criteria are defined as “Depopulated areas.”

2.4. Hospital volume

The association between hospital volume and FTR rates was analyzed. In the present study, the annual case numbers of eight representative surgical procedures were used to classify institutional departments (IDs) into the following six categories: Group A was defined as <200 cases, Group B 200–399 cases, Group C 400–599 cases, Group D 600–799 cases, Group E 800–999 cases, and Group F ≥1000 cases.

2.5. Questionnaire survey

We conducted an online questionnaire survey from July to November 2019 using the NCD system. The number of objective IDs was 2119. The 22 items in the questionnaire survey were chosen to clarify the institutional structure as well as the quality of patient care (Table S1). Then, the association between the questionnaire items and the top and bottom 20% of hospitals (in terms of FTR rates) was examined.

2.6. Statistical analysis

Descriptive statistics were calculated according to surgical volume, facility type, regional categories, and questionnaire items. The chi‐square test and Fisher's exact test were used to compare categorical data with their distributions. Cramer's V was calculated to assess the relationship between each questionnaire item and the FTR rate. Cramer's V is the most commonly used strength test for the chi‐square test. Cramer's V is a form of a correlation and is interpreted exactly the same. ¹⁴ The effect of Cramer's V is judged as low (0.1 or higher), mid (0.3 or higher), or high (0.5 or higher). The predicted mortality was calculated for each case using eight risk models previously created and reported using NCD gastroenterological data. ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² The annual numbers of cases of the eight major surgical procedures for each hospital were calculated. Multivariate logistic regression analysis was performed using 14 categorical items that were determined to be associated with operative mortality and the numbers of cases per year of the eight major surgical procedures as continuous variables. All tests were two‐sided, and values of p < 0.05 were considered statistically significant. All analyses were performed using IBM SPSS version 26 (IBM Corp.).

3. RESULTS

Among 2119 IDs registered in the NCD, 1083 (51%) responded to the survey. In this study, because eight major surgical procedures were targeted, 568 IDs (213 382 cases) without missing data for these procedures (one institution with missing data was excluded) were included in the analysis. Among these 568 institutions, 546 (55.9%) of the 976 training institutions were certified by the JSGS. Regarding the locality of the hospitals, 284 were metropolitan, 244 provincial, and 40 depopulated.

3.1. FTR rates

The overall morbidity, operative mortality, and FTR rates for all institutions were 13.3% (26 364/213 382), 2.2% (4675/213 382), and 17.7% (4675/26 364), respectively.

In terms of the association between hospital volume and operative mortality, decreases were seen with increases in the number of operations (Figure 1). The operative morbidity rate remained constant independent of the hospital volume, except for the hospitals with the lowest number of cases. The FTR rate decreased as the hospital volume increased when the hospitals with the lowest number of cases were excluded.

FIGURE 1.

Relationship between the annual numbers of eight representative surgical procedures in IDs classified into six categories. Surgical mortality decreased as the number of operations increased. Surgical morbidity remained constant and independent of hospital volume, except in hospitals with the lowest number of cases (1–200/year). The FTR rate also decreased as hospital volume increased after excluding the hospitals with the lowest number of cases.

The top and bottom 20% of hospitals in terms of the FTR rate comprised 114 and 119 IDs, respectively. A histogram of FTR rates was skewed to the right, with several hospitals showing very high FTR rates (Figure S1). The numbers of patients who underwent surgery in the top and bottom 20% of hospitals in terms of the FTR rate were 24 639 and 12 687, respectively.

A comparison of background factors between patients at the top and bottom 20% of IDs in terms of the FTR rate revealed that those in the top 20% were younger, had less trouble carrying out activities of daily living, and were less likely to have American Society of Anesthesiologists Physical Status 3 or higher (Table 1). On the other hand, the top 20% of IDs had higher proportions of patients with chronic obstructive pulmonary disease and advanced cancer who received preoperative chemotherapy and radiation therapy.

TABLE 1.

Preoperative patients' background.

Characteristics	FTR		p Value
	Top 20%	Bottom 20%
Age (mean years)	69	72	<0.001 a
Sex (Male ratio)	65.1	63.2	<0.001
BMI	22.1	21.8	<0.001
ADL before 30 days (Any assistance (%))	2.5	6.8	<0.001 a
ASA PS classification >III (%)	11.1	19.1	<0.001 a
Alcohol habitually (%)	38.0	26.9	<0.001 a
Smoking within 1 year (%)	23.1	22.0	0.021
Congestive heart failure (%)	0.4	0.6	0.035
Angina pectoris (%)	1.0	1.0	0.575
Myocardial infarction past 6 months (%)	0.4	0.5	0.449
Hypertension, %	36.2	40.1	<0.001
Previous PCI, %	2.4	2.2	0.39
Past cardiac surgery %	1.0	1.0	0.484
Respiratory distress, (%)	1.2	1.3	0.355
Chronic obstructive lung disease, %	10.4	2.0	<0.001 a
Preoperative pneumonia, %	0.4	0.5	0.272
Acute renal failure within 24 hours admission, %	0.2	0.3	0.554
Hemodialysis, %	0.6	1.0	<0.001
Esophageal varices, %	0.2	0.3	0.260
Ascites within 30 days, %	2.0	3.5	<0.001
Blood coagulation abnormalities, %	6.2	4.3	<0.001
Diabetes mellitus (insulin use), %	3.9	3.6	0.101
Disseminated Cancer, %	1.6	1.6	0.912
Chronic steroid use, %	1.3	0.7	<0.001
Chemotherapy within 90 days before operation, %	15.7	5.1	<0.001 a
Radiation within 90 days before op, %	2.7	0.7	<0.001
Sepsis before operation, %	1.5	1.6	0.255
Emergency operation, %	6.0	16.2	<0.001 a

^a Cramer V > 0.1.

Operative morbidity, operative mortality, and FTR rates in the top and bottom 20% of IDs, which were calculated based on the number of cases belonging to each group, were 13.1% (3236/24 639) and 8.4% (1069/12 687) (p < 0.001), 0.52% (129/24 639) and 4.3% (547/12 687) (p < 0. 001), and 4.0% (129/3236) and 51.2% (547/1069) (calculated p < 0.001), respectively, with statistically significant differences between the top and bottom 20% for all items. Furthermore, the average FTR rates for the top and bottom 20% of IDs were 5.6% and 44.4%, respectively.

A comparison of the characteristics of the institutions and the top/bottom 20% of IDs in terms of the FTR rate revealed no significant relationship between hospital volume and category (p = 0.489) (Table 2). Some low‐volume hospitals were included in the top 20% of IDs. Also, even hospitals with a large number of surgeries per year were included in the bottom 20% of IDs. A marginal statistical relationship was found between board certification and FTR rates (p = 0.095); only five noncertified institutions were in the top 20% of institutions. In addition, no significant difference in locality was seen between institutions with a better or worse FTR rate (p = 0.430).

TABLE 2.

Association of hospital volume, certified status, locality and FTR.

Characteristics	Entire study cohort (n = 568 hosp)	Top 20%FTR (n = 114)	Bottom 20%FTR (n = 119)	p Value
	Number of hosp
Hospital volume
Group A 1–199 cases	59	19	21	0.489
Group B 200–399 cases	137	29	40
Group C 400–599 cases	143	27	29
Group D 600–799 cases	121	22	15
Group E 800–999 cases	65	12	12
Group F ≥1000 cases	43	5	2
Hospital type
Non‐certified	22	5	12	0.095
Certified	546	109	107
Locality of hospital
Metropolitan area	284	61	55	0.430
Provincial city area	244	47	54
Depopulated area	40	6	10

Note: Hospital volume is the number of annual cases of the eight procedures.

An examination of JSGS certification and locality for each of the eight major surgical procedures revealed that esophagectomies, hepatectomies, and pancreatoduodenectomies were performed more frequently in the top 20% of institutions, whereas pan‐peritonitis operations were performed more frequently in the bottom 20% of institutions (Table S2). Regarding locality, esophagectomies, hepatectomies, and pancreaticoduodenectomies were less common in depopulated areas.

3.2. Individual complications

The number of patients who experienced operative complications is shown in Table 3. Pancreatic fistula in pancreaticoduodenectomies, biliary fistula in hepatectomies, sepsis, and recurrent nerve palsy in esophagectomies showed incidence rates >10%. Several complications, including unplanned intubation, renal dysfunction, cardiac arrest, myocardial infarction, severe sepsis/septic shock, heart failure, disseminated intravascular coagulation, and liver failure, had very high FTR rates (>30%). The complications most frequently associated with operative death, with >1000 operative deaths in 3 years, were pneumonia, unplanned intubation, renal dysfunction, and severe sepsis/septic shock. Among these complications, sepsis was associated with 40.1% (1979/4845) of the operative deaths.

TABLE 3.

Incidence and failure‐to‐rescue rates in each morbidity.

Type	Number of patients	Incidence rate (%)	Mortality case (%)	Failure‐to‐rescue rates
Surgical complications
SSI (Superficial)	10 646	5.0	786	7.4
SSI (Deep)	4366	2.0	626	14.3
SSI (Organ/Space)	11 961	5.6	946	7.9
Wound dehiscence	2542	1.2	471	18.5
Anastomotic leakage	11 593	5.4	743	6.4
Pancreatic fistula (in pancreatoduodenectomy N = 19 911)
Grade A	1208	6.1	19	1.6
Grade B	3658	18.4	52	1.4
Grade C	392	2.0	84	21.4
Biliary fistula (in liver resection N = 12 564)	1342	10.64	58	4.3
Intraabdominal abscess	6324	3.0	547	8.6
Delayed gastric emptying	3026	1.4	57	1.9
Peritonitis	1702	0.80	469	27.6
Recurrent nerve palsy (in esophagectomy N = 15 313)	1937	12.6	22	1.1
Chylothrax (in esophagectomy N = 15 313)	394	2.6	11	2.8
Tracheal necrosis (in esophagectomy N = 15 313)	31	0.20	4	12.9
Gastric tube necrosis (in esophagectomy N = 15 313)	72	0.47	12	16.7
Intestinal anastomotic stenosis	1127	0.53	17	1.5
Anastomotic ulcer	300	0.14	16	5.3
Nonsurgical complications
Severe sepsis/septic shock	6167	12.8	1979	32.1
DIC	1920	0.9	880	45.8
Pneumonia	7550	3.5	1184	15.7
Unplanned intubation	3187	1.5	1127	35.4
Pulmonary embolism	489	0.23	40	8.2
Athelectasis	3670	1.7	447	12.2
Pyothrox	475	0.22	54	11.4
Renal dysfunction	4388	2.1	1507	34.3
Urinary tract infection	2142	1.0	166	7.7
Urinary dysfunction	1506	0.71	40	2.7
Cardiac arrest	968	0.45	733	75.7
Myocardial infarction	170	0.08	67	39.4
Heart failure	890	0.42	271	30.4
Deep vein thrombosis	1100	0.52	68	6.2
CNS disturbance	712	0.08	173	24.3
Peripheral nerve palsy	249	0.12	41	16.5
Intraabdominal bleeding	1270	0.6	302	23.8
Pancreatitis	393	0.18	37	9.4
Peptic ulcer	230	0.11	24	10.4
Liver failure	542	0.25	389	71.8
(in Liver resection N = 12 564)	160	1.3	104	65.0
Liver abscess	331	0.16	35	10.6
Intractable ascites	869	0.41	207	23.8
Paralytic ileus	5054	2.4	179	3.5
Mechanical ileus	1408	0.66	55	3.9
Intestinal bleeding	1078	0.5	230	21.3
Sexual dysfunction	16	0.00	1	6.3
Others	15 496	7.3	907	5.9

3.3. Questionnaire items and FTR rates

No significant differences in patient characteristics were found between the top and bottom 20% of institutions. Concerning hospital factors in the questionnaire, 17 items showed a statistical difference between the top and bottom 20% of hospitals, including the number of ICU beds, number of ICU doctors, number of certified ICU doctors, involvement of surgeons in ICU care, regular ICU conferences, rehabilitation in the ICU, preoperative oral care by dentists, management of postoperative bleeding, management of intraabdominal abscess, regular consultations with infectious disease doctors, clinical conference attendance by medical staff, and the completion of a surgical safety checklist (Table 4). Among these items, five factors showed a strong correlation with FTR rates (Cramer's V > 0.3).

TABLE 4.

Association of questionnaire items and hospital type.

Characteristics	Top 20%FTR (n = 114)	Bottom 20%FTR (n = 119)	p‐Value	Cramer V
	number of hosp/(FTR)
Q1 Number of ICU beds (always at least 1 nurse per two patients)
Q1_ICU bed(0)	37 (3.7)	58 (56.3)	0.042	0.144
Q1_ICU_bed (1–4)	7 (3.7)	8 (48.6)
Q1_ICU_bed (5–9)	33 (4.2)	32 (48.2)
Q1_ICU_bed (10–14)	18 (4.1)	13 (47.4)
Q1_ICU_bed(15–)	19 (3.7)	8 (56.9)
Q2 Number of high care unit beds (always at least 1 nurse per four patients)
Q2_HCU bed(0)	53 (3.2)	54 (52.6)	0.921	0.236
Q2_HCU_bed (1–3)	4 (0.0)	2 (80.0)
Q2_HCU_bed (4, 5)	13 (6.1)	15 (50.0)
Q2_HCU_bed (6–15)	32 (4.4)	34 (51.6)
Q2_HCU_bed(16‐‐)	12 (4.7)	14 (49.6)
Q3_Significant changes of the number of ICU beds during study period
Yes	91 (3.9)	96 (51.2)	0.871	0.018
No	23 (4.2)	23 (51.0)
Q4 Number of physicians dedicated to ICU
Q4_ICU_Dr(0)	47 (4.0)	77 (52.6)	<0.001	0.320
Q4_ICU_Dr (1)	16 (3.9)	22 (49.8)
Q4_ICU_Dr (2)	9 (3.5)	6	(52.1)
Q4_ICU_Dr (3–7)	22 (4.1)	8	(53.4)
Q4_ICU_Dr(>8)	20 (4.2)	6	(44.7)
Q5 Number of intensivists certified by the Japanese Society of Intensive Care Medicine
Q5_ICU_SprDr(0)	52 (3.9)	84 (51.9)	<0.001	0.389
Q5_ICU_SprDr (1)	19 (4.1)	18 (53.3)
Q5_ICU_SprDr (2)	17 (4.0)	8	(50.6)
Q5_ICU_SprDr (3)	6	(3.7)	5	(51.3)
Q5_ICU_SprDr(>4)	20 (4.1)	4	(37.7)
Q6_Treatment management in ICU only with surgeons
Yes	50 (4.0)	77 (49.3)	0.001	0.211
No	64 (4.0)	42 (52.9)
Q7_ Whether ICU conferences are held or not
Yes	69 (4.0)	47 (53.2)	0.001	0.011
No	45 (4.0)	72 (49.3)
Q8_Any rehabilitation department interventions in the ICU?
Yes	92 (4.1)	80 (49.5)	0.019	0.137
No	22 (3.7)	39 (59.1)
Q9_Preoperative oral care by oral surgery or dentistry
Yes	90 (4.0)	80 (50.3)	0.044	0.282
No	24 (4.1)	39 (53.1)
Q10_Postoperative oral care by oral surgery or dentistry
Yes	98 (3.8)	93 (51.6)	0.121	0.245
No	16 (5.7)	26 (50.2)
Q11_Preoperative consultation to anesthesiologists in case of co‐morbidities
Yes	106 (4.0)	106 (50.2)	0.298	0.198
No	8 (3.0)	13 (60.0)
Q12_ Are CT imaging and emergency blood tests available on holidays?
No	1 (0.0)	0 (0)	0.489	0.034
Yes	113 (4.0)	119 (51.2)
Q13‐1_ IVR for postoperative bleeding was performed by surgeons
Yes	8 (6.3)	24 (52.7)	0.004	0.212
No	106 (3.9)	95 (49.7)
Q13‐1_IVR for postoperative bleeding was performed by interventional radiologists
Yes	95 (4.0)	72 (49.9)	<0.001	0.281
No	19 (5.7)	47 (52.0)
Q13‐1_IVR for postoperative bleeding was performed by internists
Yes	16 (4.2)	13 (49.0)	0.472	0.030
No	98 (4.0)	106 (50.8)
Q13‐2_IVR can be performed anytime (24 hours a day)
Yes	75 (4.0)	61 (49.2)	0.025	0.096
No	39 (3.9)	58 (54.5)
Q14_Drainage for intraabdominal abscess was performed by surgeons
Yes	101 (4.0)	117 (51.1)	0.003	0.031
No	13 (3.1)	2 (54.5)
Q14_Drainage for intraabdominal abscess was performed by interventional radiologists
Yes	74 (4.0)	43 (46.9)	<0.001	0.437
No	40 (3.9)	76 (54.3)
Q14_Drainage for intraabdominal abscess was performed by internists, endoscopists
Yes	28 (3.9)	15 (46.3)	0.019	0.094
No	86 (4.0)	104 (52.3)
Q15_Consultations to infection control team for decision making of antimicrobial therapy
Yes	33 (3.7)	14 (50.7)	0.001	0.364
No	81 (4.2)	105 (51.2)
Q16_Sepsis (including septic DIC) treatment strategy_ usage of drugs with weak evidence
Yes	106 (3.9)	117 (50.8)	0.055	0.130
No	8	(4.6)	2 (59.5)
Q16_ Sepsis (including septic DIC) treatment strategy_ usage of drugs with weak evidence; Heparin
Yes	72 (3.8)	63 (49.9)	0.114	0.098
No	42 (4.1)	56 (51.9)
Q16_ Sepsis (including septic DIC) treatment strategy_ usage of drugs with weak evidence; protease inhibitors
Yes	74 (3.9)	76 (51.4)	0.868	0.022
No	40 (4.0)	43 (50.0)
Q16_ Sepsis (including septic DIC) treatment strategy_ usage of drugs with weak evidence; AT‐III
Yes	94 (4.0)	93 (50.9)	0.402	0.074
No	20 (3.5)	26 (50.6)
Q16_ Sepsis (including septic DIC) treatment strategy_ usage of drugs with weak evidence; rTM
Yes	99 (4.0)	106 (50.4)	0.600	0.194
No	15 (3.2)	13 (54.1)
Q16_ Sepsis (including septic DIC) treatment strategy_ usage of drugs with weak evidence; IVIG
Yes	86 (4.1)	90 (49.9)	0.973	0.043
No	28 (3.4)	29 (53.3)
Q16_ Sepsis (including septic DIC) treatment strategy_ usage of drugs with weak evidence; low dose corticosteroids
Yes	59 (4.2)	58 (50.6)	0.645	0.074
No	55 (3.5)	61 (51.1)
Q17_The following treatments are available for patients with postoperative complications
PMX
Yes	104 (4.0)	110 (50.2)	0.736	0.058
No	10 (4.3)	9 (80.0)
CHDF
Yes	106 (4.0)	112 (50.6)	0.724	0.082
No	8 (2.5)	7 (66.7)
ExtraCorporeal Membrane Oxygenation
Yes	57 (4.0)	52 (49.5)	0.335	0.152
No	57 (4.0)	67 (52.8)
Plama Exchange
Yes	85 (4.2)	85 (50.6)	0.590	0.134
No	29 (3.4)	34 (51.4)
Q18_Patiant care by individual attending physician (or by surgical team)
Yes	57 (3.6)	71 (51.0)	0.138	0.225
No	57 (4.2)	48 (51.4)
Q19_ Patient care on holidays is provided by individual attending physicians.
Yes	71 (3.9)	80 (52.9)	0.429	0.056
No	43 (4.3)	39 (48.4)
Q20_regular M&M conference
Yes	74 (3.9)	75 (49.8)	0.764	0.245
No	40 (4.2)	44 (52.9)
Q21_Medical staffs participate in the decision making conferences
Yes	60 (4.1)	87 (51.1)	0.001	0.234
No	54 (3.9)	32 (51.5)
Q22_Implementation of surgical check list
Yes	110 (4.0)	116 (51.2)	0.717	0.058
No	4	(0.0)	3 (50.0)
Q22 Implementation of surgical check list_self‐introduction
Yes	74 (3.9)	60 (50.4)	0.025	0.307
No	40 (4.3)	59 (51.9)
Q22 Implementation of surgical check list_patients' information
Yes	108 (3.9)	116 (51.2)	0.325	0.093
No	6 (6.3)	3 (0)
Q22 Implementation of surgical check list_operative procedures
Yes	109 (4.0)	116 (51.2)	0.492	0.049
No	5	(4.2)	3 (0)
Q22 Implementation of surgical check list_Anticipated major events
Yes	91 (4.0)	81 (51.5)	0.041	0.275
No	23 (4.4)	38 (50.5)

3.4. Appropriate cutoff value for the number of certified ICU doctors

Next, the crude mortality rate, risk‐adjusted mortality rate, and observed number of deaths in relation to the expected number of deaths (O:E ratio) by number of certified ICU specialists were examined (Figure 2). Based on the O:E ratio curve, we decided to classify the ICUs into two groups: those with zero or one certified intensivist and those with two or more certified intensivists.

FIGURE 2.

Crude mortality and O:E ratio for the procedures by the number of board‐certified intensivists. The O:E ratios were calculated by dividing the observed operative mortality by the risk‐adjusted mortality. Based on the O:E ratio curve, it seemed appropriate to divide the institutions into two groups: those with zero or one certified intensivist and those with two or more certified intensivists.

3.5. Multivariate logistic analysis for FTR and operative mortality

Of the 17 factors extracted in Table 4, Q1 and Q4 were excluded because they were correlated with Q5, and Q8 was excluded because no difference in mortality was seen between the top and bottom 20% of institutions. The outcome for the FTR rates was defined as the top 20% of FTR rates in the multivariate logistic analysis. The results show that three factors were significant in the multivariate analysis of the top 20% of FTR rates: average number of surgeries per year, medical staff participation in conferences, and completion of a surgical safety checklist (Table S3). Then, six items were found to be associated with operative mortality in the multivariate logistic analysis, including two or more certified intensivists, preoperative oral care, management of postoperative bleeding, management of intraabdominal abscess, regular consultations with infectious disease doctors, and clinical conference attendance by medical staff (Table 5).

TABLE 5.

Multivariable logistic regression analysis for operative mortality (case‐basis).

Questionnaire item/mortality patients/number of ope patients		Operative mortality (%)	p‐Value	Exp(B)	EXP(B) 95% CI
					Lower	Upper
Average number of cases a year			0.001	0.999	0.998	0.999
Q5_ICU_Certified ICU Dr
0.1	2993/123 594	2.4	0.016	0.899	0.824	0.980
≧2	1682/89 788	1.9
Q6_ICU manage by surgical team alone
Yes	2017/81 483	2.5	0.494	0.969	0.885	1.061
No	2658/131 899	2.0
Q7_ Whether ICU conferences are held or not
Yes	2870/135 404	2.1	0.131	0.937	0.862	1.019
No	1805/77 978	2.3
Q9_Preoperative oral care
Yes	3876/184 966	2.1	0.000	0.844	0.771	0.923
No	799/28 416	2.8
Q13_1_ IVR for postoperative bleeding was performed by surgeons
Yes	686/24 096	2.8	0.000	1.213	1.093	1.347
No	3989/189 286	2.1
Q13_1_ IVR for postoperative bleeding was performed by interventional radiologists
Yes	3829/184 473	2.1	0.082	0.903	0.805	1.013
No	846/28 909	2.9
Q13_2 IVR can be performed anytime (24 hours a day)
Yes	3633/168769	2.1	0.307	1.047	0.958	1.144
No	1042/44613	2.3
Q14 Drainage for intraabdominal abscess was performed by surgeons
Yes	4351/194 557	2.2	0.006	1.209	1.057	1.383
No	324/18 825	1.7
Q14 Drainage for intraabdominal abscess was performed by interventional radiologists
Yes	2753/139 694	2.0	0.056	0.923	0.85	1.002
No	1922/73 688	2.6
Q14 Drainage for intraabdominal abscess was performed by internists, endoscopists
Yes	1128/55 744	2.0	0.990	1.000	0.925	1.083
No	3547/157 638	2.3
Q15 Consultations to infection control team for decision making of antimicrobial therapy
Yes	1311/72 232	1.8	0.021	0.911	0.842	0.986
No	3364/141 150	2.4
Q18 Is the patient assigned to an individual surgeon (YES) or a surgical team (NO)?
Yes	2461/107 048	2.3	0.719	1.012	0.946	1.083
No	2214/106 334	2.1
Q20_ Are mortality & morbidity conferences held on a regular basis?
Yes	3282/153 392	2.1	0.412	0.969	0.90	1.044
No	1393/59 990	2.3
Q21_Participation of Medical staff in the clinical conference
Yes	3161/135 504	2.3	0.007	1.104	1.027	1.186
No	1514/77 878	1.9
Q22_Implementation of surgical check list_self introduction
Yes	3072/149 276	2.2	0.578	0.910	0.654	1.267
No	1603/64 106	2.4

The factor of clinical conference attendance by medical staff was excluded because it was oddly inversely related to operative death, leaving only 50 IDs that met all five factors (50/568 = 8.8%). On a case‐count basis, 36 270 surgeries (17.0% of the entire cohort) were performed at institutions that met the five criteria. FTR rates were better with increases in the number of matched factors (Table 6).

TABLE 6.

Number of matched factors and FTR.

Number of factor	Number of institutions	Mean FTR (%)
0	42	29
1	79	27
2	108	24
3	169	19
4	120	18
5	50	13

4. DISCUSSION

Because operative mortality occurs after major postoperative complications in most cases, the ability of hospitals to help patients recover from postoperative complications is important to help guarantee safe surgical procedures. Therefore, the achievement of a lower (better) FTR rate is a reliable indicator of a good hospital. The present study revealed that in addition to hospital volume, several human factors are associated with low operative mortality and FTR rates after eight major gastrointestinal surgeries, such as having two or more intensivists, providing preoperative oral care, managing postoperative bleeding, managing intraabdominal abscess, and holding regular consultations with infectious disease doctors, based on data from a large‐scale nationwide database in Japan. Furthermore, <10% of the institutions that responded to the survey met the criteria for a good hospital.

Hospital volume is well known to be associated with surgical mortality for various types of surgery. ² , ³ , ²³ In the present study, there was no relationship between the number of cases and the incidence of postoperative complications (Figure 1). In general, surgical mortality is an extension of postoperative complications; therefore, institutions with higher rates of postoperative complications are often thought to have higher rates of surgical mortality. Several recent large database studies have shown that the incidence of postoperative complications does not correlate with surgical mortality. For example, in a previous study by Miura et al., ²⁴ the postoperative complication rate after hepatectomy was lower in non‐ board‐certified centers with a low annual number of hepatobiliary pancreatic operations (23.5%), and the complication rate was higher in high‐volume, board‐certified centers (28.0%). Furthermore, our previous study on hepatopancreatoduodenectomy showed no difference in operative mortality between centers for small hepatectomies, but a large difference in operative mortality and FTR rates for large hepatectomies. ²⁵ This may be because high‐volume centers operate on more difficult cases and have higher postoperative complication rates. We speculated that the incidence of postoperative complications may be higher in high‐volume centers because they operate on more difficult cases. The complexity of each procedure could not be evaluated in the present study. However, we did find a background factor of a significantly higher rate of preoperative chemotherapy in the top 20% of hospitals (Table 1). This may suggest that the top 20% IDs perform a greater proportion of highly complex surgeries on patients with advanced disease. We believe that such institutions have a low operative mortality rate because they are better equipped and staffed to help patients recover even if the postoperative complication rate is high. Thus, highly complex surgeries have high complication rates and should be performed at hospitals with good FTR rates. In the future, not only the number of cases but also the FTR rate should be a quality indicator of hospital function in Japan.

It is a misconception that if the FTR rate is good, there is no need to worry about postoperative complications. This is because an FTR rate of zero, i.e., zero mortality from serious postoperative complications, is extremely difficult if not impossible to achieve. In fact, in this study, even the top 20% of hospitals had an average FTR rate of 5.6%. On the other hand, some hospitals with low postoperative complication rates are unable to rescue patients once a complication occurs. In the present study, one of the characteristics of the bottom 20% IDs was the high number of emergency surgeries (Table 1, bottom line). Such hospitals have a low postoperative complication rate because they usually perform less difficult surgeries, which are less likely to cause postoperative complications. However, regarding emergency surgeries, these bottom 20% IDs are forced to operate even on patients in poor general condition because they are unable to refer the patients to more suitable facilities. In such cases, once a complication occurs, it cannot be resolved because there are no ICU specialists or interventional radiology (IVR) specialists on staff. Therefore, such facilities need to collaborate with nearby high‐volume centers in case of postoperative complications.

Several investigators have reported that nonsurgical physician factors may be related to improvements in FTR rates. For example, Pronovost et al. ²⁶ reported that adequate staffing was associated with lower hospital mortality, with a pooled estimate of the relative risk of hospital mortality of 0.71. Ward et al. ¹⁰ reported that hospitals with lower FTR rates tended to employ more advanced practice providers, such as physician assistants and nurse practitioners. Hospitals with a higher percentage of nurses with a baccalaureate or advanced degree have reported lower mortality and life‐saving rates for patients undergoing surgical procedures. ²⁷ Presumably, the advantages of the top 20% of institutions are expected to include early detection rates and the integrated, coordinated, and intensive treatment of postoperative complications owing to the general abundance of human resources.

ICU care managed by intensivists has been reported to be associated with a significant reduction in post‐trauma in‐hospital mortality. ²⁸ Ward et al. ¹⁰ used the Michigan Surgical Quality Collaborative database to compare 52 hospitals in Michigan. The percentage of board‐certified intensivists was 88% in hospitals with a low FTR rate versus 60% in hospitals with a high FTR rate. In the present study, having two or more certified intensivists was associated with a better FTR rate. In fact, the present study found that 37.7% (43/114) of the top 20% of institutions met the requirement of having at least two certified intensivists, whereas only 14.3% (17/119) of the bottom 20% of institutions fulfilled this requirement (Table 4). On the other hand, polymyxin B immobilized fiber cartridge (110/119), continuous hemodiafiltration (112/119), plasma exchange (85/119), and extracorporeal membrane oxygenation (52/119) can be performed in most of the bottom 20% of institutions. Therefore, the availability of such state‐of‐the‐art devices is not directly linked to FTR rates.

Interventional radiology has become the preferred method of managing postoperative bleeding because it is less invasive and has a lower mortality rate. Several studies, including systematic reviews, have reported significantly lower mortality rates for IVR when compared to surgical treatment. ²⁹ In this study, 83.3% of the top 20% of hospitals were equipped to treat intra‐abdominal bleeding by interventional radiologists. In contrast, approximately 40% (47/119) of the bottom 20% of hospitals did not have an IVR specialist available. Similarly, 64.9% (74/114) of the top 20% of hospitals were able to perform intra‐abdominal abscesses drainage by interventional radiologists, while 63.9% (76/119) of the bottom 20% of hospitals were unable to do so. This may be due to a shortage of IVR specialists and the existence of regional disparities in the distribution of specialists, which prevents smaller hospitals from hiring IVR specialists for less frequently occurring serious complications.

The large difference in FTR rates between the top and bottom 20% of institutions is a problem in Japan. In the present study, the top 20% of institutions had an average FTR rate of 5.6%, which is equivalent to that in the United States, ⁸ , ¹⁰ whereas the bottom 20% of institutions had an extremely poor average FTR rate of 44.4%, compared to 16.7% in the worst 20% of hospitals in the US study. On a case‐count basis, FTR rates of the top and bottom 20% of institutions were 4.0% (129/3236) and 51.2% (547/1069), respectively (relative risk = 25.3). About half of the bottom 20% of hospitals (58/119) had no ICU beds, and 77/119 had surgeons managing the ICU. Surgeons become increasingly busy when having to manage postoperative complications, which increases the likelihood of exhaustion and burnout. When more surgeons are on the verge of burnout, medical errors increase and patient safety might not be maintained. ³⁰ Therefore, we should aim to improve the environment surrounding surgeons in the bottom 20% of institutions through policy recommendations to fulfill mandatory conditions. It might be possible to prevent surgeon burnout by promoting appropriate functional sharing among hospitals, and this could serve as a driving force to facilitate an acceptable work–life balance among surgeons.

In general, hospitals in depopulated areas have been estimated to have worse postoperative outcomes because they are often low‐volume hospitals. Takahashi et al. ¹³ reported that low‐volume hospitals account for 38.4% of esophageal cancer surgeries in depopulated areas, and only 5.7% in metropolitan areas. However, in the present study, locality did not affect FTR rates. It may be speculated that hospitals in depopulated areas already share the function of performing surgeries with low FTR rates exclusively and refer those with high FTR rates to large metropolitan tertiary hospitals. Indeed, in this study, fewer esophagectomies, hepatectomies, and pancreatoduodenectomies were performed in depopulated areas. Therefore, it seems that locality does not affect surgical outcomes among hospitals that have good cooperation. ³¹

This study has several limitations. First, this was a survey of only eight major surgeries, so the results may not be generalizable to other procedures. However, these major surgeries representing digestive surgery are recognized as surrogate indicators. Second, the reason that the participation of medical staff was significantly higher in the bottom 20% of institutions cannot be explained. However, it is difficult to imagine a direct relationship between the participation of medical staff in clinical conferences and operative mortality. Therefore, some unknown factors need to be investigated further.

5. CONCLUSION

The results of the present study revealed that several hospital factors surrounding surgical treatment, as characterized by abundant human resources, are closely related to better postoperative recovery from severe complications. Therefore, for the further improvement of surgical care, these factors should be included in future board certification and institutional accreditation systems.

AUTHOR CONTRIBUTIONS

IE, AT, HT, and YS made substantial contributions to the conceptualization and design of the work and acquisition of the data. IE and AT contributed to the writing and preparation of the original draft. TK, YH, and RM prepared documents and assisted with statistical calculations for the Ethics Committee. All authors critically reviewed the manuscript and approved the final version for submission.

FUNDING INFORMATION

This research did not receive any specific grant from funding agencies in the public, commercial, or nonprofit organizations.

CONFLICT OF INTEREST STATEMENT

AT, HT, and HM are affiliated with the Department of Healthcare Quality Assessment at The University of Tokyo, which is a social collaboration department supported by the National Clinical Database, Johnson & Johnson KK, and Nipro Corporation. The other authors declare no conflicts of interest for this article. The authors (Itaru Endo, Yohsihiro Kakeji, Yuko Kitagawa, and Yasuyuki Seto) are currently editorial members of AGS and will not be involved in the review of this paper or in the selection of reviewers.

ETHICS STATEMENT

The protocol for this research project was approved by a suitably constituted Ethics Committee of the institution and conforms to the provisions of the Declaration of Helsinki (Institutional Review Board of Yokohama City University, Approval nos. B190400030 and B200806086). Informed consent was provided at each institution participating in the National Clinical Database.

Supporting information

Figure S1

Table S1

Table S2

Table S3

Endo I, Takahashi A, Tachimori H, Miyata H, Homma Y, Kumamoto T, et al. Requirements for hospitals in Japan to have low operative mortality and failure‐to‐rescue rates. Ann Gastroenterol Surg. 2024;8:342–355. 10.1002/ags3.12745