- Email: [email protected]
The effect of ethnicity on in-hospital mortality following emergency abdominal surgery: a national cohort study using Hospital Episode Statistics
p u b l i c h e a l t h 1 2 9 ( 2 0 1 5 ) 1 4 9 6 e1 5 0 2
Available online at www.sciencedirect.com
Public Health journal homepage: www.elsevier.com/puhe
Original Research
The effect of ethnicity on in-hospital mortality following emergency abdominal surgery: a national cohort study using Hospital Episode Statistics R.S. Vohra a, F. Evison b, I. Bejaj b, D. Ray b, P. Patel a, T.D. Pinkney a,* a
Academic Department of Surgery, School of Cancer Sciences, University of Birmingham, Birmingham, B15 2TH, UK Department of Informatics, University Hospitals Birmingham NHS Foundation Trust, Queen Elizabeth Hospital, Birmingham, B15 2TH, UK
b
article info
abstract
Article history:
Objectives: Ethnicity has complex effects on health and the delivery of health care in part
Received 7 January 2015
related to language and cultural barriers. This may be important in patients requiring
Received in revised form
emergency abdominal surgery where delays have profound impact on outcomes. The aim
30 May 2015
here was to test if variations in outcomes (e.g. in-hospital mortality) exist by ethnic group
Accepted 16 July 2015
following emergency abdominal surgery.
Available online 28 August 2015
Study design: Retrospective cohort study using population-level routinely collected administrative data from England (Hospital Episode Statistics).
Keywords:
Methods: Adult patients undergoing emergency abdominal operations between April 2008
Emergency surgery
and March 2012 were identified. Operations were divided into: ‘major’, ‘hepatobiliary’ or
Routinely collected data
‘appendectomy/minor’. The primary outcome was all cause in-hospital mortality. Uni-
Surgical outcomes
variable and multivariable analysis odds ratios (OR with 95% confidence intervals, CI) adjusting for selected factors were performed. Results: 359,917 patients were identified and 80.7% of patients were White British, 4.7% White (Other), 2.4% Afro-Caribbean, 1.6% Indian, 2.6% Chinese, 3.1% Asian (Other) and 4.9% not known, with crude in-hospital mortality rates of 4.4%, 3.1%, 2.0%, 2.6%, 1.6%, 1.7% and 5.17%, respectively. The majority of patients underwent appendectomy/minor (61.9%) compared to major (20.9%) or hepatobiliary (17.2%) operations (P < 0.001) with an in-hospital mortality of 1.7%, 11.5% and 3.9% respectively. Adjusted mortality was largely similar across ethnic groups except where ethnicity was not recorded (compared to White British patients following major surgery OR 2.05, 95% 1.82e2.31, P < 0.01, hepatobiliary surgery OR 2.78, 95% CI 2.31e3.36, P ¼ 0.01 and appendectomy/minor surgery OR 1.78, 95% 1.52e2.08, P < 0.01). Conclusions: Ethnicity is not associated with poorer outcomes following emergency abdominal surgery. However, ethnicity is not recorded in 5% of this cohort and this represents an important, yet un-definable, group with significantly poorer outcomes. © 2015 The Royal Society for Public Health. Published by Elsevier Ltd. All rights reserved.
* Corresponding author. Tel.: þ44 (0) 121 414 6924; fax: þ44 (0) 121 472 1230. E-mail address: [email protected] (T.D. Pinkney). http://dx.doi.org/10.1016/j.puhe.2015.07.038 0033-3506/© 2015 The Royal Society for Public Health. Published by Elsevier Ltd. All rights reserved.
p u b l i c h e a l t h 1 2 9 ( 2 0 1 5 ) 1 4 9 6 e1 5 0 2
Introduction Ethnicity can be classified by referring to a community of people who share the same culture and/or by referring to an ancestral population which comprises their self-identity.1 Self-reported ethnicity captures both the shared experiences/culture of an individual and their self-identity. According to the 2011 Consensus in some areas of the United Kingdom, especially around London, over 70% of the population report their ethnicity as ‘non-white’.2 Even those who are considered ‘White’ are comprised of a heterogeneous group of residents from Poland, Greece, Romania, Bulgaria and other Eastern countries. Ethnicity has complex effects on health and the delivery of health care.3e5 This is a major issue not only in the UK, but also in other countries such as mainland Europe and the United States, both formed of similarly diverse communities.6 Many challenges exist when treating patients from an ethnic background different to that of the health service provider. These can be patient-related including language barriers and cultural barriers e.g. omissions of sensitive elements in the history and adequate expose patients for examinations.7,8 In addition, there are known provider-related biases linked to racial and ethnic prejudice.9 At a population level, the effects of these factors are difficult to quantify individually, but may have a cumulative effect in delaying diagnosis, treatment and outcomes. Rapid diagnosis and treatment is essential in patients requiring urgent and emergency abdominal surgery.10e12 Inappropriate delays produce profound effects on short-term outcomes such as in-hospital mortality, which are widely used as markers of quality.13,14 Inappropriate delays in care resulting from issues arising from ethnicity would be expected to have a significant impact on those requiring emergency surgery. However, this is unclear. In England, Hospital Episode Statistics (HES) is an is an administrative dataset that collates information on all National Health Service (NHS) and private patients admitted to NHS hospitals in England on a per-episode basis. HES can monitor population-level outcomes following elective and emergency operations. Self-reported ethnicity is recorded in HES. HES data in recent years have high completeness of ethnic group information (typically exceeding 90%) and accurate in 95% of records when validated.15e17 There have been notable improvements in ethnicity recording in the past decade. The aim of this study was to test if variations in outcomes (e.g. in-hospital mortality and length of hospital stay) exist by ethnic group in patients following emergency abdominal surgery.
Methods Hospital Episode Statistics (HES) A description of the HES database has been published previously.18 In brief, it is an administrative dataset that collates information on all NHS and private patients admitted to NHS hospitals in England. Each admission contains a primary diagnosis and secondary diagnoses which are categorised according to ICD-10 (international classification of diseases, 10th revision),19 along with patient-level demographic data
1497
including age, ethnicity and home postal code. This study is exempt from UK National Research Ethics Committee approval as it involved analysis of an existing dataset of anonymized data for service evaluation. A data sharing agreement with the Health and Social Care Information Centre (HSCIC) to use Hospital Episode Statistics data has been granted. Studies performed at the Department of Informatics, Queen Elizabeth Hospital Birmingham fall under that arrangement and has been approved by Data Access Advisory Group (DAAG) hosted by the HSCIC.
Database inclusions and variable coding All patients over the age of 18 years classified as an ‘emergency’ admission between 1st April 2008 and 31st March 2012 and undergoing a surgical procedure on either their digestive tract or an abdominal organ were included. Supplementary Table 1 shows the Office of Population Censuses and Surveys Classification of Interventions and Procedures (OPCS-4.5) codes used to identify patients. OPCS-4 codes were grouped into: major, hepatobilary and minor/appendicectomy. By using the unique patient identifier it was possible to determine whether a patient had had at least one emergency admission within England in the year prior to the admission of interest. Ethnicity is recorded as part of the general demographics. Ethnicity is grouped into White British; White Irish or Other White; Black African, Caribbean, Other or Mixed; Indian; Pakistani, Bangladeshi, Other Asian or Mixed; Chinese, Other or Other Mixed; or Not stated/known. Diagnoses were recoded into broad categories using the ICD-10 code in the primary diagnosis field; digestive system, neoplasm or other. Patients were grouped into six age cohorts: 18e34, 35e44, 45e54, 55e64, 65e74, and 75 þ years according to age at time of surgery. The adjusted Charlson co-morbidity score was considered in three categories: 0, 1e4, and 5.20 The indices of multiple deprivation (IMD) 2007 are an overall score of deprivation derived from seven domains, based on routinely collected statistics; income, employment, health, education, training and skills, barriers to housing and services, crime and living environment.21 Quintiles derived from these scores, classified from 1 (most deprived) to 5 (least deprived) were used here.
Outcome variables The primary endpoint of interest was in-hospital all-cause postoperative mortality. This was information is recorded within HES. ‘Length of stay’ is the time (in days) spent in hospital during the index admission and ‘days until operation’ is the time (in days) between the date of admission and the operation taking place. Medians (with interquartile range) are referred to in unadjusted analyses.
Univariable and multivariable analysis Categorical variables were tested for association with the death outcome variables using the c2 test, the ManneWhitney test was used to test non-normally distributed continuous variables. Variables including age, gender, primary diagnosis, co-morbidities, deprivation, and previous
1498
p u b l i c h e a l t h 1 2 9 ( 2 0 1 5 ) 1 4 9 6 e1 5 0 2
emergency admissions were considered in the multilevel logistic model, which was first fitted with all significant variables and then tested to see whether any variables could be dropped, hospital providers included as the second level; thus admissions were nested within provider.
Statistical analysis Statistical analyses were conducted in SAS V.9.2 (SAS Institute, Cary, North Carolina, USA). P-values of <0.010 were considered statistically significant. The PROC GLIMMIX procedure was used to perform the multilevel analyses.
Results General demographics During the period of study 359,917 adult patients underwent emergency abdominal operations in England. Demographics are shown in Table 1. There was a fall in the percentage of people in which ethnicity was ‘not stated or known’ from 6.0% to 4.7% over the period studied, but still represented the
second largest group (17,563 patients; 4.9% of overall cohort) after White British patients. The percentage of admissions where ethnicity was not recorded by region of residence in England is shown in Supplementary Fig. 1. The demographics of each major ethnic group were analysed (Table 2). When the demographics of patients with a ‘not stated or known’ ethnicity status were compared to other ethnic groups there were similarities to some groups, but differences to others. Patients where ethnicity was ‘not stated or known’ were significantly younger (median age 39 [IQR: 26e56] vs. 50 [32e70], P < 0.001), had undergone minor surgery more frequently (76.1% vs. 59.4%, P < 0.001), had fewer comorbidities (Charlson score ¼ 0, 82.5% vs 68.3% P < 0.001) and were less likely to have had a previous admission (9.9% vs 33.4%, P < 0.001) compared to white British patients. However, deprivation scores and diagnosis were similar in both groups. Age ranges and diagnosis were similar in the ‘not stated or known’ group compared to those identified as Black, Indian, Pakistani, Bangladeshi and Chinese groups. Fewer comorbidities (Charlson score ¼ 0, 82.5% vs. 77.3%, P < 0.001) and minor operations (76.1% vs. 70.3%, P < 0.001) occurred in the ‘not stated or known’ group compared to the Black, Indian, Pakistani, Bangladeshi and Chinese groups.
Table 1 e General demographics. Demographic
n (%)
Gender
Male Female
191,350 (53.2) 168,463 (46.8)
Age group (years)
18e34 35e44 45e54 55e64 65e74 75þ
112,812 52,064 47,166 44,701 43,512 59,558
Co-morbidity score
0 1e4 5þ
252,403 (70.2) 43,212 (12.0) 64,198 (17.8)
Deprivation
1 e most deprived 2 3 4 5 e least deprived
Ethnicity
White British White Irish or Other White Black African, Black Caribbean, Other Black, Mixed Indian Pakistani, Bangladeshi, Other Asian, Mixed Chinese, Other or Other Mixed Not stated/known
Year of admission
2008/09 2009/10 2010/11 2011/12
Procedures
Major Hepatobilary Minor/appendectomy
75,373 (21.0) 61,862 (17.2) 222,578 (61.9)
Diagnosis
Benign Neoplasm Other Previous emergency admission 1 year prior
251,891 32,448 75,474 113,505
(31.4) (14.5) (13.1) (12.4) (12.1) (16.6)
82,591 75,840 71,169 66,796 63,417
(23.0) (21.1) (19.8) (18.7) (17.6)
290,359 16,982 8760 5867 11,116 9166 17,563
(80.7) (4.7) (2.4) (1.6) (3.1) (2.6) (4.9)
84,712 87,852 92,112 95,137
(23.5) (24.4) (25.6) (26.4)
(70.0) (9.0) (21.0) (31.6)
1499
p u b l i c h e a l t h 1 2 9 ( 2 0 1 5 ) 1 4 9 6 e1 5 0 2
Table 2 e General demographics by ethnic group [n (%)]. Demographic
White British
Gender
Male Female
Age group (years)
18e34 35e44 45e54 55e64 65e74 75þ
Co-morbidity score
0
150,200 (51.7) 140,159 (48.3) 82,141 39,674 38,182 38,506 38,267 53,589
(28.3) (13.7) (13.2) (13.3) (13.2) (18.5)
White Irish or other white
Indian
Pakistani, Bangladeshi, other Asian, mixed
9415 (55.4) 7567 (44.6)
4697 (53.6) 4063 (46.4)
3599 (61.3) 2268 (38.7)
6860 (61.7) 4256 (38.3)
5588 (61.0) 10,991 (62.6) 3578 (39.0) 6572 (37.4)
6831 2620 1947 1742 1605 2237
3629 1890 1451 537 661 592
2499 1031 746 656 482 453
5821 2125 1213 793 650 514
4507 1748 1147 720 525 519
(40.2) (15.4) (11.5) (10.3) (9.5) (13.2)
198,373 (68.3) 12,768 (75.2)
(41.4) (21.6) (16.6) (6.1) (7.6) (6.8)
(42.6) (17.6) (12.7) (11.2) (8.2) (7.7)
(52.4) (19.1) (10.9) (7.1) (5.9) (4.6)
6554 (74.8)
4253 (72.5)
8538 (76.8)
772 (13.2) 842 (14.4)
1320 (11.9) 1258 (11.3)
Chinese, other or other mixed
Not stated/ known
Black African, Black Caribbean, other Black, mixed
(49.2) (19.1) (12.5) (7.9) (5.7) (5.7)
7384 2976 2480 1747 1322 1654
(42.0) (16.9) (14.1) (10.0) (7.5) (9.4)
7428 (81.0) 14,489 (82.5)
1e4 5þ
36,297 (12.5) 55,689 (19.2)
1532 (9.0) 2682 (15.8)
1008 (11.5) 1198 (13.7)
Deprivation
1 e most deprived 2 3 4 5 e least deprived
59,605 58,873 59,160 57,608 55,113
(20.5) (20.3) (20.4) (19.8) (19.0)
4784 4267 3187 2423 2321
(28.2) (25.1) (18.8) (14.3) (13.7)
4330 2509 1047 533 341
(49.4) (28.6) (12.0) (6.1) (3.9)
1724 1707 1141 726 569
(29.4) (29.1) (19.5) (12.4) (9.7)
5458 2596 1542 910 610
(49.1) (23.4) (13.9) (8.2) (5.5)
3263 2426 1558 1009 910
(35.6) (26.6) (17.0) (11.0) (9.9)
3427 3462 3534 3587 3553
(19.5) (19.7) (20.1) (20.4) (20.2)
Year of admission
2008/09 2009/10 2010/11 2011/12
67,963 71,461 74,342 76,593
(23.4) (24.6) (25.6) (26.4)
3972 3856 4437 4717
(23.4) (22.7) (26.1) (27.8)
1971 2071 2358 2360
(22.5) (23.6) (26.9) (26.9)
1300 1446 1587 1534
(22.2) (24.7) (27.1) (26.2)
2390 2763 2975 2988
(21.5) (24.9) (26.8) (26.9)
2057 2223 2380 2506
(22.4) (24.3) (26.0) (27.3)
5059 4032 4033 4439
(28.8) (23.0) (23.0) (25.3)
Procedures
Major 65,654 (22.6) 3138 (18.5) Hepatobilary 52,273 (18.0) 2675 (15.8) Minor/appendectomy 17,2432 (59.4) 11,169 (65.8)
1376 (15.7) 1247 (14.2) 6137 (70.1)
694 (11.8) 815 (13.9) 4358 (74.3)
1031 (9.3) 1757 (15.8) 8328 (74.9)
1159 (12.6) 2321 (13.2) 1219 (13.3) 1876 (10.7) 6788 (74.1) 13,366 (76.1)
Diagnosis
Benign Neoplasm Other
6023 (68.8) 5,69 (6.5) 2168 (24.8)
4160 (70.9) 280 (4.8) 1427 (24.3)
7811 (70.3) 431 (3.9) 2874 (25.9)
6580 (71.8) 1,2940 (73.7) 460 (5.0) 1037 (5.9) 2126 (23.2) 3586 (20.4)
2738 (31.3)
1810 (30.9)
3275 (29.5)
2242 (24.5)
Previous emergency admission 1 year prior
20,2413 (69.7) 1,1964 (70.5) 28,396 (9.8) 1275 (7.5) 59,550 (20.5) 3743 (22.0) 96,957 (33.4)
4753 (28.0)
791 (8.6) 947 (10.3)
1492 (8.5) 1582 (9.0)
1730 (9.9)
Timing of surgery and length of hospital stay
In-hospital mortality
The median length between admission and the surgical procedure was one day (IQR 0e2). The median length of overall hospital stay was significantly shorter in those patients where ethnicity was ‘not stated or known’ compared to either White British (two days [IQR 1e5] vs. four days [2e9], P < 0.001). When the three separate operative groups were considered, the median length of hospital stay in patients who were discharged was 11 days (IQR 6e21) following major procedures, seven days (IQR 4e13) after hepatobilary procedures and two days (IQR 1e4 days) following minor/appendectomy. White British patients were more likely to have a longer time period between admission and undergoing an appendectomy/ minor surgery compared to those in the ‘not stated or known’ group (one day [IQR 0e1] vs. 0 days [IQR 0e1], P < 0.001). Again, in patients undergoing hepatobiliary and major surgery, the total length of hospital stay was shorter for those whose ethnicity was ‘not stated or known’ compared to White British patients (six days [IQR 3e10] vs. seven days [IQR 4e13], P < 0.001 and nine days [IQR 5e17] vs. 11 days [IQR 6e21], P < 0.001, respectively).
The in-hospital mortality following appendectomy/minor, major and hepatobiliary operations were 1.7%, 11.5% and 3.9%, respectively. Tables 3 and 4 show the crude and adjusted
Table 3 e Crude rates of death within hospital by ethnic group and surgery type [n (%)]. Ethnic group
Major
White British 7557 (11.5) White Irish or Other 278 (8.9) White Black African, 77 (5.6) Caribbean, Other, mixed Indian 71 (10.2) Pakistani, 60 (5.8) Bangladeshi, Other Asian, Mixed Chinese, Other, 70 (6.0) Other Mixed Not stated/known 548 (23.6)
Hepatobiliary
Minor/ appendectomy
1996 (3.8) 85 (3.2)
3140 (1.8) 161 (1.4)
50 (4.0)
52 (0.9)
32 (3.9) 62 (3.5)
52 (1.2) 55 (0.7)
38 (3.1)
43 (0.6)
153 (8.2)
207 (1.6)
1500
p u b l i c h e a l t h 1 2 9 ( 2 0 1 5 ) 1 4 9 6 e1 5 0 2
Table 4 e Odds ratios for death within hospital following hepatobiliary, major or minor/appendectomy [OR (95% CI)]. Demographic
Hepatobiliary
Major
Minor/appendectomy
Gender
Male Female
1.11 (1.02, 1.21) P ¼ 0.02 1
0.94 (0.89, 1.00) P ¼ 0.04 0.89 (0.83, 0.96) P ¼ 0.01 1 1
Age group (years)
18e29 30e44 45e64 65þ
0.12 (0.08, 0.17) P < 0.01 0.20 (0.17, 0.25) P < 0.01 0.39 (0.35, 0.43) P < 0.01 1
0.04 (0.03, 0.06) P < 0.01 0.11 (0.10, 0.14) P < 0.01 0.33 (0.31, 0.35) P < 0.01 1
0.01 (0.01, 0.01) P < 0.01 0.12 (0.10, 0.14) P < 0.01 0.37 (0.34, 0.41) P < 0.01 1
Co-morbidity score
0 1e4 5þ
1 1.35 (1.15, 1.59) P < 0.01 4.13 (3.70, 4.59) P < 0.01
1 1.68 (1.55, 1.81) P < 0.01 3.27 (3.09, 3.45) P < 0.01
1 1.48 (1.30, 1.69) P < 0.01 6.19 (5.68, 6.74) P < 0.01
Deprivation
1 e most deprived 2 3 4 5 e least deprived
1.37 1.26 1.15 1.02 1
Year of Admission
2008/09 2009/10 2010/11 2011/12
1.15 (1.02, 1.30) P ¼ 0.02 1.07 (0.94, 1.20) P ¼ 0.30 1.01 (0.89, 1.14) P ¼ 0.92 1
1.33 (1.24, 1.44) P < 0.01 1.35 (1.22, 1.49) P < 0.01 1.16 (1.07, 1.25) P ¼ 0.01 1.23 (1.11, 1.36) P < 0.01 1.03 (0.96, 1.11) P ¼ 0.43 1.11 (1.00, 1.22) P ¼ 0.05 1 1
Diagnosis
Benign Neoplasms Other
0.32 (0.28, 0.36) P < 0.01 1.46 (1.28, 1.65) P < 0.01 1
1.13 (1.03, 1.24) P ¼ 0.01 0.55 (0.50, 0.59) P < 0.01 1.08 (0.98, 1.20) P ¼ 0.13 2.17 (1.93, 2.45) P < 0.01 1 1
Ethnicity
White British White Irish or Other White Black African, Caribbean, Other, mixed Indian Pakistani, Bangladeshi, Other Asian, Mixed Chinese, Other, Other mixed Not stated/known
1 0.79 (0.62, 1.00) P ¼ 0.05 1.16 (0.85, 1.59) P ¼ 0.35
1 1 0.84 (0.72, 0.97) P ¼ 0.02 0.99 (0.83, 1.19) P ¼ 0.94 0.60 (0.46, 0.79) P < 0.01 0.67 (0.50, 0.90) P ¼ 0.01
0.91 (0.62, 1.35) P ¼ 0.64 1.11 (0.84, 1.48) P ¼ 0.47
1.17 (0.86, 1.59) P ¼ 0.32 0.84 (0.62, 1.16) P ¼ 0.29 0.72 (0.52, 0.98) P ¼ 0.04 0.53 (0.39, 0.71) P < 0.01
0.97 (0.68, 1.37) P ¼ 0.86 2.78 (2.31, 3.36) P ¼ 0.01
0.71 (0.54, 0.94) P ¼ 0.02 0.72 (0.52, 0.99) P ¼ 0.05 2.05 (1.82, 2.31) P < 0.01 1.78 (1.52, 2.08) P < 0.01
0.06 (0.016)
0.04 (0.007)
Error variance (standard deviation)
(1.18, 1.58) (1.09, 1.46) (0.99, 1.32) (0.88, 1.18)
P P P P
< 0.01 ¼ 0.02 ¼ 0.06 ¼ 0.802
1.44 (1.32, 1.58) 1.22 (1.12, 1.34) 1.17 (1.07, 1.27) 1.04 (0.96, 1.14) 1
P P P P
< 0.01 < 0.01 < 0.01 ¼ 0.36
1.35 1.12 1.10 1.07 1
(1.20, 1.52) (0.99, 1.26) (0.98, 1.24) (0.94, 1.20)
P < 0.01 P ¼ 0.08 P ¼ 0.12 P ¼ 0.30
0.05 (0.013)
Each variable was adjusted for the other variables in the table to calculate the odds ratio.
in-hospital mortality following each operative group analysed by ethnic group, respectively. Age, deprivation, co-morbidities and year of surgery were all independent risk factors for inpatient mortality following all three types of surgery. In the three types of surgeries studied, the intra-provider correlation coefficients ranged from 1 to 5%. When compared to White British patients, adjusted in-hospital mortality was significantly higher across all surgical types if ethnicity was ‘not stated or known’. Certain ethnic groups such as AfroCaribbean and Chinese patients had lower adjusted inhospital mortality following major surgery compared to White British patients.
Discussion Variations in-hospital mortality following surgery is important to patients and health service providers. There are a growing number of factors implicated including surgeon caseload, hospital volume and the day of the week operating.22e25 This study found no evidence of detrimental outcomes for any named ethnic group. This would suggest that there are no identifiable barriers (e.g. cultural, linguistic or other) that impact upon the provision of emergency surgical care. The
major caveat to this is the identification of a significant proportion of patients where ethnicity is ‘not stated or known’. This group comprises around 5% of the total cohort, but represents an important, yet un-definable, cohort with significantly poorer outcomes compared to others. The 5% missing data rate for the ethnicity has improved compared to historic estimates of HES;14 this group still represents the second group here. An attempt was made to further reduce missing data by scanning individual patientlevel data for other admissions either before or after the index admission to elucidate ethnicity data that may have been recorded on other occasions. Name recognition software has been used by others but is currently only 7% sensitive26 and individual names are not available on the HES database. The ‘not stated or known’ cohort appeared to have some similarities to certain ethnic groups and differences to others. It is likely to represent a mix of different ethnicities rather than one single ethnic group. This aside it is interesting that the proportion of patients missing ethnicity data varied significantly according to geographical region (Supplementary Fig. 1). The reasons for this are not clear. Unrecorded ethnicity was associated with disproportionately worse in-hospital mortality compared to the rest of the population studied. This is despite a shorter time between
p u b l i c h e a l t h 1 2 9 ( 2 0 1 5 ) 1 4 9 6 e1 5 0 2
admission and an operation in selected emergency surgeries and overall length of stay. One hypothesis for the missing ethnicity data is that the group represents an ‘un-well’ set of patients that require rapid operations. However, an important caveat is this cohort is discharged quickest; this would not be expected if patients had undergone emergency surgery of this type. Another possibility is that these patients come from a setting of enhanced social support and can thereby be discharged earlier, but this would not necessarily account for the earlier operations. There are weaknesses in analysing administrative datasets including continued systematic reporting errors, missed cases and inaccuracies despite improvements in quality assurance.27,28 There is a lack of clinical information including disease severity and general health within the HES dataset such that there are other unknown variables that will not have been fully controlled for. In addition, there are established provider-related biases linked to racial and ethnic prejudice which could not be controlled for here.9 Ethnicity-related variability in both access to surgery and outcomes after surgery identified by others in the elective surgical setting29,30 was not replicated in this cohort of patients for whom ethnicity was recorded. In addition, from the cohort presented here, the variability in the data between providers was low. Taken together this suggests that in the emergency setting the central tenet of the NHS to provide good health care across providers holds true. The group of patients for whom ethnicity is not recorded in part impacts upon the power of this study. The worse outcomes in this group and the increasing ethnic diversity of England mean a strategic endeavour to improve awareness and coding of ethnicity is vital as we appear to be missing and perhaps underserving a high risk group.
Author statements Acknowledgements The authors would like to thank attendees at the Association of Surgeons of Great Britain and Ireland 2014 for their comments.
Ethical approval This study is exempt from UK National Research Ethics Committee approval as it involved analysis of an existing dataset of anonymized data for service evaluation. A data sharing agreement with the Health and Social Care Information Centre (HSCIC) to use Hospital Episode Statistics data has been granted. Studies performed at the Department of Informatics, Queen Elizabeth Hospital Birmingham fall under that arrangement and has been approved by Data Access Advisory Group (DAAG) hosted by the HSCIC.
Funding None.
1501
Competing interests None.
Authors' contributions RSV, TP and FE participated in the conception, design, analysis, writing and editing of this study. All authors participated in the design and writing of the manuscript. FE, IB, DR and PP participated in the statistical analysis. RSV, FE, IB and DR are the guarantors. All authors read and approved the final manuscript.
Appendix A. Supplementary data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.puhe.2015.07.038.
references
1. Isajiw WW. Definition and dimensions of ethnicity: a theoretical framework. In: Challenges of measuring an ethnic World: Science, politics and reality: Proceedings of the Joint CanadaUnited States Conference on the measurement of ethnicity. Washington, DC: U.S. Government Printing Office; April 1e3, 1992. p. 407e27. 2. http://www.ons.gov.uk/ons/rel/census/2011-census/keystatistics-for-local-authorities-in-england-and-wales/index. html [Accessed 01.06.2014]. 3. Levine DA, Neidecker MV, Kiefe CI, Karve S, Williams LS, Allison JJ. Racial/ethnic disparities in access to physician care and medications among US stroke survivors. Neurology 2011;76:53e61. 4. Hassan MO, Arthurs Z, Sohn VY, Steele SR. Race does not impact colorectal cancer treatment or outcomes with equal access. Am J Surg 2009;197:485e90. 5. Fleury AC, Ibeanu OA, Bristow RE. Racial disparities in surgical care for uterine cancer. Gynecol Oncol 2011;121:571e6. 6. Trivedi AN, Grebla RC, Wright SM, Washington DL. Despite improved quality of care in the veterans affairs health system, racial disparity persists for important clinical outcomes. Health Aff (Millwood) 2011;30:707e15. Dao M, Durieux-Paillard S. Quality in 7. Hudelson P, Dominice practice: integrating routine collection of patient language data into hospital practice. Int J Qual Health Care 2013;25(4):437e42. 8. Juckett G. Cross-cultural medicine. Am Fam Physician 2005 Dec 1;72(11):2267e74. 9. Bhogal R. Spectre of racism in health and health care: lessons from history and the United States. BMJ 1998;316:1970. 10. NCEPOD. Emergency admissions: a journey in the right direction? A report of the national confidential enquiry into patient outcome and death. London: NCEPOD; 2007. 11. The Royal College of Surgeons of England. Emergency surgery: standards for unscheduled care. RCSENG e professional standards and regulation, London; 2011. 12. Kluger Y, Ben-Ishay O, Sartelli M, Ansaloni L, Abbas AE, Agresta F, et al. World society of emergency surgery study group initiative on timing of acute care surgery classification (TACS). World J Emerg Surg 2013;8:17.
1502
p u b l i c h e a l t h 1 2 9 ( 2 0 1 5 ) 1 4 9 6 e1 5 0 2
13. Huang Q, Thind A, Dreyer JF, Zaric GF. The impact of delays to admission from the emergency department on inpatient outcomes. BMC Emerg Med 2010;10:16. 14. Ghaferi AA, Birkmeyer JD, Dimick JB. Variation in hospital mortality associated with inpatient surgery. N Engl J Med 2009;361:1368e75. 15. The NHS Information Centre. How good is HES ethnic coding and where do the problems lie? [accessed 28.04.15], http://www. hesonline.nhs.uk/Ease/servlet/ContentServer? siteID¼1937&categoryID¼171; 2011. 16. Jack RH, Linklater KM, Hofman D. Ethnicity coding in a regional cancer registry and in hospital episode Statistics. BMC Public Health 2006;6:281. 17. Saunders CL, Abel GA, El Turabi A, Ahmed F, Lyratzopoulos G. Accuracy of routinely recorded ethnic group information compared with self-reported ethnicity: evidence from the English Cancer patient experience survey. BMJ Open 2013;3. e002882. 18. Burns EM, Bottle A, Aylin P, Darzi A, Nicholls RJ, Faiz O. Variation in reoperation after colorectal surgery in England as an indicator of surgical performance: retrospective analysis of hospital episode statistics. Br Med J 2011;343. d4836. 19. International classification of diseases, 10th revision. World Health Organization; 2010 [Retrieved February 26, 2013]. 20. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987;40:373e83. 21. Carstairs V, Morris R. Deprivation: explaining differences in mortality between Scotland and England and Wales. BMJ 1989;299:886e9. 22. Morris EJA, Taylor EF, Thomas JD, Quirke P, Finan PJ, Coleman MP, et al. Thirty-day postoperative mortality
23.
24.
25.
26.
27.
28.
29.
30.
after colorectal cancer surgery in England. Gut 2011;60(6):806e13. Aylin P, Alexandrescu R, Jen MH, Mayer EK, Bottle A. Day of week of procedure and 30 day mortality for elective surgery: retrospective analysis of hospital episode statistics. BMJ 2013;346. f2424. Bach PB, Cramer LD, Schrag D, Downey RJ, Gelfand SE, Begg CB. The influence of hospital volume on survival after resection for lung cancer. N Engl J Med 2001;345:181e8. Finks JF, Osborne NH, Birkmeyer JD. Trends in hospital volume and operative mortality for high-risk surgery. N Engl J Med 2011;364:2128e37. Ryan R, Vernon S, Lawrence G, Wilson S. Use of name recognition software, census data and multiple imputation to predict missing data on ethnicity: application to cancer registry records. BMC Med Inform Decis Mak 2012;12:3. Sinha S, Peach G, Poloniecki JD, Thompson MM, Holt PJ. Studies using English administrative data (hospital episode statistics) to assess health-care outcomesdsystematic review and recommendations for reporting. Eur J Public Health 2013;23(1):86e92. Westaby S, Archer N, Manning N, Adwani S, Grebenik C, Ormerod O, et al. Comparison of hospital episode statistics and central cardiac audit database in public reporting of congenital heart surgery mortality. BMJ 2007;335(7623):759. Skinner J, Weinstein JN, Sporer SM, Wennberg JE. Racial, ethnic, and geographic disparities in rates of knee arthroplasty among medicare patients. N Engl J Med 2003;349:1350e9. Liu JH, Zingmond DS, McGory ML, SooHoo NF, Ettner SL, Brook RH, et al. Disparities in the utilization of high-volume hospitals for complex surgery. JAMA 2006;296(16):1973e80.
Available online at www.sciencedirect.com
Public Health journal homepage: www.elsevier.com/puhe
Original Research
The effect of ethnicity on in-hospital mortality following emergency abdominal surgery: a national cohort study using Hospital Episode Statistics R.S. Vohra a, F. Evison b, I. Bejaj b, D. Ray b, P. Patel a, T.D. Pinkney a,* a
Academic Department of Surgery, School of Cancer Sciences, University of Birmingham, Birmingham, B15 2TH, UK Department of Informatics, University Hospitals Birmingham NHS Foundation Trust, Queen Elizabeth Hospital, Birmingham, B15 2TH, UK
b
article info
abstract
Article history:
Objectives: Ethnicity has complex effects on health and the delivery of health care in part
Received 7 January 2015
related to language and cultural barriers. This may be important in patients requiring
Received in revised form
emergency abdominal surgery where delays have profound impact on outcomes. The aim
30 May 2015
here was to test if variations in outcomes (e.g. in-hospital mortality) exist by ethnic group
Accepted 16 July 2015
following emergency abdominal surgery.
Available online 28 August 2015
Study design: Retrospective cohort study using population-level routinely collected administrative data from England (Hospital Episode Statistics).
Keywords:
Methods: Adult patients undergoing emergency abdominal operations between April 2008
Emergency surgery
and March 2012 were identified. Operations were divided into: ‘major’, ‘hepatobiliary’ or
Routinely collected data
‘appendectomy/minor’. The primary outcome was all cause in-hospital mortality. Uni-
Surgical outcomes
variable and multivariable analysis odds ratios (OR with 95% confidence intervals, CI) adjusting for selected factors were performed. Results: 359,917 patients were identified and 80.7% of patients were White British, 4.7% White (Other), 2.4% Afro-Caribbean, 1.6% Indian, 2.6% Chinese, 3.1% Asian (Other) and 4.9% not known, with crude in-hospital mortality rates of 4.4%, 3.1%, 2.0%, 2.6%, 1.6%, 1.7% and 5.17%, respectively. The majority of patients underwent appendectomy/minor (61.9%) compared to major (20.9%) or hepatobiliary (17.2%) operations (P < 0.001) with an in-hospital mortality of 1.7%, 11.5% and 3.9% respectively. Adjusted mortality was largely similar across ethnic groups except where ethnicity was not recorded (compared to White British patients following major surgery OR 2.05, 95% 1.82e2.31, P < 0.01, hepatobiliary surgery OR 2.78, 95% CI 2.31e3.36, P ¼ 0.01 and appendectomy/minor surgery OR 1.78, 95% 1.52e2.08, P < 0.01). Conclusions: Ethnicity is not associated with poorer outcomes following emergency abdominal surgery. However, ethnicity is not recorded in 5% of this cohort and this represents an important, yet un-definable, group with significantly poorer outcomes. © 2015 The Royal Society for Public Health. Published by Elsevier Ltd. All rights reserved.
* Corresponding author. Tel.: þ44 (0) 121 414 6924; fax: þ44 (0) 121 472 1230. E-mail address: [email protected] (T.D. Pinkney). http://dx.doi.org/10.1016/j.puhe.2015.07.038 0033-3506/© 2015 The Royal Society for Public Health. Published by Elsevier Ltd. All rights reserved.
p u b l i c h e a l t h 1 2 9 ( 2 0 1 5 ) 1 4 9 6 e1 5 0 2
Introduction Ethnicity can be classified by referring to a community of people who share the same culture and/or by referring to an ancestral population which comprises their self-identity.1 Self-reported ethnicity captures both the shared experiences/culture of an individual and their self-identity. According to the 2011 Consensus in some areas of the United Kingdom, especially around London, over 70% of the population report their ethnicity as ‘non-white’.2 Even those who are considered ‘White’ are comprised of a heterogeneous group of residents from Poland, Greece, Romania, Bulgaria and other Eastern countries. Ethnicity has complex effects on health and the delivery of health care.3e5 This is a major issue not only in the UK, but also in other countries such as mainland Europe and the United States, both formed of similarly diverse communities.6 Many challenges exist when treating patients from an ethnic background different to that of the health service provider. These can be patient-related including language barriers and cultural barriers e.g. omissions of sensitive elements in the history and adequate expose patients for examinations.7,8 In addition, there are known provider-related biases linked to racial and ethnic prejudice.9 At a population level, the effects of these factors are difficult to quantify individually, but may have a cumulative effect in delaying diagnosis, treatment and outcomes. Rapid diagnosis and treatment is essential in patients requiring urgent and emergency abdominal surgery.10e12 Inappropriate delays produce profound effects on short-term outcomes such as in-hospital mortality, which are widely used as markers of quality.13,14 Inappropriate delays in care resulting from issues arising from ethnicity would be expected to have a significant impact on those requiring emergency surgery. However, this is unclear. In England, Hospital Episode Statistics (HES) is an is an administrative dataset that collates information on all National Health Service (NHS) and private patients admitted to NHS hospitals in England on a per-episode basis. HES can monitor population-level outcomes following elective and emergency operations. Self-reported ethnicity is recorded in HES. HES data in recent years have high completeness of ethnic group information (typically exceeding 90%) and accurate in 95% of records when validated.15e17 There have been notable improvements in ethnicity recording in the past decade. The aim of this study was to test if variations in outcomes (e.g. in-hospital mortality and length of hospital stay) exist by ethnic group in patients following emergency abdominal surgery.
Methods Hospital Episode Statistics (HES) A description of the HES database has been published previously.18 In brief, it is an administrative dataset that collates information on all NHS and private patients admitted to NHS hospitals in England. Each admission contains a primary diagnosis and secondary diagnoses which are categorised according to ICD-10 (international classification of diseases, 10th revision),19 along with patient-level demographic data
1497
including age, ethnicity and home postal code. This study is exempt from UK National Research Ethics Committee approval as it involved analysis of an existing dataset of anonymized data for service evaluation. A data sharing agreement with the Health and Social Care Information Centre (HSCIC) to use Hospital Episode Statistics data has been granted. Studies performed at the Department of Informatics, Queen Elizabeth Hospital Birmingham fall under that arrangement and has been approved by Data Access Advisory Group (DAAG) hosted by the HSCIC.
Database inclusions and variable coding All patients over the age of 18 years classified as an ‘emergency’ admission between 1st April 2008 and 31st March 2012 and undergoing a surgical procedure on either their digestive tract or an abdominal organ were included. Supplementary Table 1 shows the Office of Population Censuses and Surveys Classification of Interventions and Procedures (OPCS-4.5) codes used to identify patients. OPCS-4 codes were grouped into: major, hepatobilary and minor/appendicectomy. By using the unique patient identifier it was possible to determine whether a patient had had at least one emergency admission within England in the year prior to the admission of interest. Ethnicity is recorded as part of the general demographics. Ethnicity is grouped into White British; White Irish or Other White; Black African, Caribbean, Other or Mixed; Indian; Pakistani, Bangladeshi, Other Asian or Mixed; Chinese, Other or Other Mixed; or Not stated/known. Diagnoses were recoded into broad categories using the ICD-10 code in the primary diagnosis field; digestive system, neoplasm or other. Patients were grouped into six age cohorts: 18e34, 35e44, 45e54, 55e64, 65e74, and 75 þ years according to age at time of surgery. The adjusted Charlson co-morbidity score was considered in three categories: 0, 1e4, and 5.20 The indices of multiple deprivation (IMD) 2007 are an overall score of deprivation derived from seven domains, based on routinely collected statistics; income, employment, health, education, training and skills, barriers to housing and services, crime and living environment.21 Quintiles derived from these scores, classified from 1 (most deprived) to 5 (least deprived) were used here.
Outcome variables The primary endpoint of interest was in-hospital all-cause postoperative mortality. This was information is recorded within HES. ‘Length of stay’ is the time (in days) spent in hospital during the index admission and ‘days until operation’ is the time (in days) between the date of admission and the operation taking place. Medians (with interquartile range) are referred to in unadjusted analyses.
Univariable and multivariable analysis Categorical variables were tested for association with the death outcome variables using the c2 test, the ManneWhitney test was used to test non-normally distributed continuous variables. Variables including age, gender, primary diagnosis, co-morbidities, deprivation, and previous
1498
p u b l i c h e a l t h 1 2 9 ( 2 0 1 5 ) 1 4 9 6 e1 5 0 2
emergency admissions were considered in the multilevel logistic model, which was first fitted with all significant variables and then tested to see whether any variables could be dropped, hospital providers included as the second level; thus admissions were nested within provider.
Statistical analysis Statistical analyses were conducted in SAS V.9.2 (SAS Institute, Cary, North Carolina, USA). P-values of <0.010 were considered statistically significant. The PROC GLIMMIX procedure was used to perform the multilevel analyses.
Results General demographics During the period of study 359,917 adult patients underwent emergency abdominal operations in England. Demographics are shown in Table 1. There was a fall in the percentage of people in which ethnicity was ‘not stated or known’ from 6.0% to 4.7% over the period studied, but still represented the
second largest group (17,563 patients; 4.9% of overall cohort) after White British patients. The percentage of admissions where ethnicity was not recorded by region of residence in England is shown in Supplementary Fig. 1. The demographics of each major ethnic group were analysed (Table 2). When the demographics of patients with a ‘not stated or known’ ethnicity status were compared to other ethnic groups there were similarities to some groups, but differences to others. Patients where ethnicity was ‘not stated or known’ were significantly younger (median age 39 [IQR: 26e56] vs. 50 [32e70], P < 0.001), had undergone minor surgery more frequently (76.1% vs. 59.4%, P < 0.001), had fewer comorbidities (Charlson score ¼ 0, 82.5% vs 68.3% P < 0.001) and were less likely to have had a previous admission (9.9% vs 33.4%, P < 0.001) compared to white British patients. However, deprivation scores and diagnosis were similar in both groups. Age ranges and diagnosis were similar in the ‘not stated or known’ group compared to those identified as Black, Indian, Pakistani, Bangladeshi and Chinese groups. Fewer comorbidities (Charlson score ¼ 0, 82.5% vs. 77.3%, P < 0.001) and minor operations (76.1% vs. 70.3%, P < 0.001) occurred in the ‘not stated or known’ group compared to the Black, Indian, Pakistani, Bangladeshi and Chinese groups.
Table 1 e General demographics. Demographic
n (%)
Gender
Male Female
191,350 (53.2) 168,463 (46.8)
Age group (years)
18e34 35e44 45e54 55e64 65e74 75þ
112,812 52,064 47,166 44,701 43,512 59,558
Co-morbidity score
0 1e4 5þ
252,403 (70.2) 43,212 (12.0) 64,198 (17.8)
Deprivation
1 e most deprived 2 3 4 5 e least deprived
Ethnicity
White British White Irish or Other White Black African, Black Caribbean, Other Black, Mixed Indian Pakistani, Bangladeshi, Other Asian, Mixed Chinese, Other or Other Mixed Not stated/known
Year of admission
2008/09 2009/10 2010/11 2011/12
Procedures
Major Hepatobilary Minor/appendectomy
75,373 (21.0) 61,862 (17.2) 222,578 (61.9)
Diagnosis
Benign Neoplasm Other Previous emergency admission 1 year prior
251,891 32,448 75,474 113,505
(31.4) (14.5) (13.1) (12.4) (12.1) (16.6)
82,591 75,840 71,169 66,796 63,417
(23.0) (21.1) (19.8) (18.7) (17.6)
290,359 16,982 8760 5867 11,116 9166 17,563
(80.7) (4.7) (2.4) (1.6) (3.1) (2.6) (4.9)
84,712 87,852 92,112 95,137
(23.5) (24.4) (25.6) (26.4)
(70.0) (9.0) (21.0) (31.6)
1499
p u b l i c h e a l t h 1 2 9 ( 2 0 1 5 ) 1 4 9 6 e1 5 0 2
Table 2 e General demographics by ethnic group [n (%)]. Demographic
White British
Gender
Male Female
Age group (years)
18e34 35e44 45e54 55e64 65e74 75þ
Co-morbidity score
0
150,200 (51.7) 140,159 (48.3) 82,141 39,674 38,182 38,506 38,267 53,589
(28.3) (13.7) (13.2) (13.3) (13.2) (18.5)
White Irish or other white
Indian
Pakistani, Bangladeshi, other Asian, mixed
9415 (55.4) 7567 (44.6)
4697 (53.6) 4063 (46.4)
3599 (61.3) 2268 (38.7)
6860 (61.7) 4256 (38.3)
5588 (61.0) 10,991 (62.6) 3578 (39.0) 6572 (37.4)
6831 2620 1947 1742 1605 2237
3629 1890 1451 537 661 592
2499 1031 746 656 482 453
5821 2125 1213 793 650 514
4507 1748 1147 720 525 519
(40.2) (15.4) (11.5) (10.3) (9.5) (13.2)
198,373 (68.3) 12,768 (75.2)
(41.4) (21.6) (16.6) (6.1) (7.6) (6.8)
(42.6) (17.6) (12.7) (11.2) (8.2) (7.7)
(52.4) (19.1) (10.9) (7.1) (5.9) (4.6)
6554 (74.8)
4253 (72.5)
8538 (76.8)
772 (13.2) 842 (14.4)
1320 (11.9) 1258 (11.3)
Chinese, other or other mixed
Not stated/ known
Black African, Black Caribbean, other Black, mixed
(49.2) (19.1) (12.5) (7.9) (5.7) (5.7)
7384 2976 2480 1747 1322 1654
(42.0) (16.9) (14.1) (10.0) (7.5) (9.4)
7428 (81.0) 14,489 (82.5)
1e4 5þ
36,297 (12.5) 55,689 (19.2)
1532 (9.0) 2682 (15.8)
1008 (11.5) 1198 (13.7)
Deprivation
1 e most deprived 2 3 4 5 e least deprived
59,605 58,873 59,160 57,608 55,113
(20.5) (20.3) (20.4) (19.8) (19.0)
4784 4267 3187 2423 2321
(28.2) (25.1) (18.8) (14.3) (13.7)
4330 2509 1047 533 341
(49.4) (28.6) (12.0) (6.1) (3.9)
1724 1707 1141 726 569
(29.4) (29.1) (19.5) (12.4) (9.7)
5458 2596 1542 910 610
(49.1) (23.4) (13.9) (8.2) (5.5)
3263 2426 1558 1009 910
(35.6) (26.6) (17.0) (11.0) (9.9)
3427 3462 3534 3587 3553
(19.5) (19.7) (20.1) (20.4) (20.2)
Year of admission
2008/09 2009/10 2010/11 2011/12
67,963 71,461 74,342 76,593
(23.4) (24.6) (25.6) (26.4)
3972 3856 4437 4717
(23.4) (22.7) (26.1) (27.8)
1971 2071 2358 2360
(22.5) (23.6) (26.9) (26.9)
1300 1446 1587 1534
(22.2) (24.7) (27.1) (26.2)
2390 2763 2975 2988
(21.5) (24.9) (26.8) (26.9)
2057 2223 2380 2506
(22.4) (24.3) (26.0) (27.3)
5059 4032 4033 4439
(28.8) (23.0) (23.0) (25.3)
Procedures
Major 65,654 (22.6) 3138 (18.5) Hepatobilary 52,273 (18.0) 2675 (15.8) Minor/appendectomy 17,2432 (59.4) 11,169 (65.8)
1376 (15.7) 1247 (14.2) 6137 (70.1)
694 (11.8) 815 (13.9) 4358 (74.3)
1031 (9.3) 1757 (15.8) 8328 (74.9)
1159 (12.6) 2321 (13.2) 1219 (13.3) 1876 (10.7) 6788 (74.1) 13,366 (76.1)
Diagnosis
Benign Neoplasm Other
6023 (68.8) 5,69 (6.5) 2168 (24.8)
4160 (70.9) 280 (4.8) 1427 (24.3)
7811 (70.3) 431 (3.9) 2874 (25.9)
6580 (71.8) 1,2940 (73.7) 460 (5.0) 1037 (5.9) 2126 (23.2) 3586 (20.4)
2738 (31.3)
1810 (30.9)
3275 (29.5)
2242 (24.5)
Previous emergency admission 1 year prior
20,2413 (69.7) 1,1964 (70.5) 28,396 (9.8) 1275 (7.5) 59,550 (20.5) 3743 (22.0) 96,957 (33.4)
4753 (28.0)
791 (8.6) 947 (10.3)
1492 (8.5) 1582 (9.0)
1730 (9.9)
Timing of surgery and length of hospital stay
In-hospital mortality
The median length between admission and the surgical procedure was one day (IQR 0e2). The median length of overall hospital stay was significantly shorter in those patients where ethnicity was ‘not stated or known’ compared to either White British (two days [IQR 1e5] vs. four days [2e9], P < 0.001). When the three separate operative groups were considered, the median length of hospital stay in patients who were discharged was 11 days (IQR 6e21) following major procedures, seven days (IQR 4e13) after hepatobilary procedures and two days (IQR 1e4 days) following minor/appendectomy. White British patients were more likely to have a longer time period between admission and undergoing an appendectomy/ minor surgery compared to those in the ‘not stated or known’ group (one day [IQR 0e1] vs. 0 days [IQR 0e1], P < 0.001). Again, in patients undergoing hepatobiliary and major surgery, the total length of hospital stay was shorter for those whose ethnicity was ‘not stated or known’ compared to White British patients (six days [IQR 3e10] vs. seven days [IQR 4e13], P < 0.001 and nine days [IQR 5e17] vs. 11 days [IQR 6e21], P < 0.001, respectively).
The in-hospital mortality following appendectomy/minor, major and hepatobiliary operations were 1.7%, 11.5% and 3.9%, respectively. Tables 3 and 4 show the crude and adjusted
Table 3 e Crude rates of death within hospital by ethnic group and surgery type [n (%)]. Ethnic group
Major
White British 7557 (11.5) White Irish or Other 278 (8.9) White Black African, 77 (5.6) Caribbean, Other, mixed Indian 71 (10.2) Pakistani, 60 (5.8) Bangladeshi, Other Asian, Mixed Chinese, Other, 70 (6.0) Other Mixed Not stated/known 548 (23.6)
Hepatobiliary
Minor/ appendectomy
1996 (3.8) 85 (3.2)
3140 (1.8) 161 (1.4)
50 (4.0)
52 (0.9)
32 (3.9) 62 (3.5)
52 (1.2) 55 (0.7)
38 (3.1)
43 (0.6)
153 (8.2)
207 (1.6)
1500
p u b l i c h e a l t h 1 2 9 ( 2 0 1 5 ) 1 4 9 6 e1 5 0 2
Table 4 e Odds ratios for death within hospital following hepatobiliary, major or minor/appendectomy [OR (95% CI)]. Demographic
Hepatobiliary
Major
Minor/appendectomy
Gender
Male Female
1.11 (1.02, 1.21) P ¼ 0.02 1
0.94 (0.89, 1.00) P ¼ 0.04 0.89 (0.83, 0.96) P ¼ 0.01 1 1
Age group (years)
18e29 30e44 45e64 65þ
0.12 (0.08, 0.17) P < 0.01 0.20 (0.17, 0.25) P < 0.01 0.39 (0.35, 0.43) P < 0.01 1
0.04 (0.03, 0.06) P < 0.01 0.11 (0.10, 0.14) P < 0.01 0.33 (0.31, 0.35) P < 0.01 1
0.01 (0.01, 0.01) P < 0.01 0.12 (0.10, 0.14) P < 0.01 0.37 (0.34, 0.41) P < 0.01 1
Co-morbidity score
0 1e4 5þ
1 1.35 (1.15, 1.59) P < 0.01 4.13 (3.70, 4.59) P < 0.01
1 1.68 (1.55, 1.81) P < 0.01 3.27 (3.09, 3.45) P < 0.01
1 1.48 (1.30, 1.69) P < 0.01 6.19 (5.68, 6.74) P < 0.01
Deprivation
1 e most deprived 2 3 4 5 e least deprived
1.37 1.26 1.15 1.02 1
Year of Admission
2008/09 2009/10 2010/11 2011/12
1.15 (1.02, 1.30) P ¼ 0.02 1.07 (0.94, 1.20) P ¼ 0.30 1.01 (0.89, 1.14) P ¼ 0.92 1
1.33 (1.24, 1.44) P < 0.01 1.35 (1.22, 1.49) P < 0.01 1.16 (1.07, 1.25) P ¼ 0.01 1.23 (1.11, 1.36) P < 0.01 1.03 (0.96, 1.11) P ¼ 0.43 1.11 (1.00, 1.22) P ¼ 0.05 1 1
Diagnosis
Benign Neoplasms Other
0.32 (0.28, 0.36) P < 0.01 1.46 (1.28, 1.65) P < 0.01 1
1.13 (1.03, 1.24) P ¼ 0.01 0.55 (0.50, 0.59) P < 0.01 1.08 (0.98, 1.20) P ¼ 0.13 2.17 (1.93, 2.45) P < 0.01 1 1
Ethnicity
White British White Irish or Other White Black African, Caribbean, Other, mixed Indian Pakistani, Bangladeshi, Other Asian, Mixed Chinese, Other, Other mixed Not stated/known
1 0.79 (0.62, 1.00) P ¼ 0.05 1.16 (0.85, 1.59) P ¼ 0.35
1 1 0.84 (0.72, 0.97) P ¼ 0.02 0.99 (0.83, 1.19) P ¼ 0.94 0.60 (0.46, 0.79) P < 0.01 0.67 (0.50, 0.90) P ¼ 0.01
0.91 (0.62, 1.35) P ¼ 0.64 1.11 (0.84, 1.48) P ¼ 0.47
1.17 (0.86, 1.59) P ¼ 0.32 0.84 (0.62, 1.16) P ¼ 0.29 0.72 (0.52, 0.98) P ¼ 0.04 0.53 (0.39, 0.71) P < 0.01
0.97 (0.68, 1.37) P ¼ 0.86 2.78 (2.31, 3.36) P ¼ 0.01
0.71 (0.54, 0.94) P ¼ 0.02 0.72 (0.52, 0.99) P ¼ 0.05 2.05 (1.82, 2.31) P < 0.01 1.78 (1.52, 2.08) P < 0.01
0.06 (0.016)
0.04 (0.007)
Error variance (standard deviation)
(1.18, 1.58) (1.09, 1.46) (0.99, 1.32) (0.88, 1.18)
P P P P
< 0.01 ¼ 0.02 ¼ 0.06 ¼ 0.802
1.44 (1.32, 1.58) 1.22 (1.12, 1.34) 1.17 (1.07, 1.27) 1.04 (0.96, 1.14) 1
P P P P
< 0.01 < 0.01 < 0.01 ¼ 0.36
1.35 1.12 1.10 1.07 1
(1.20, 1.52) (0.99, 1.26) (0.98, 1.24) (0.94, 1.20)
P < 0.01 P ¼ 0.08 P ¼ 0.12 P ¼ 0.30
0.05 (0.013)
Each variable was adjusted for the other variables in the table to calculate the odds ratio.
in-hospital mortality following each operative group analysed by ethnic group, respectively. Age, deprivation, co-morbidities and year of surgery were all independent risk factors for inpatient mortality following all three types of surgery. In the three types of surgeries studied, the intra-provider correlation coefficients ranged from 1 to 5%. When compared to White British patients, adjusted in-hospital mortality was significantly higher across all surgical types if ethnicity was ‘not stated or known’. Certain ethnic groups such as AfroCaribbean and Chinese patients had lower adjusted inhospital mortality following major surgery compared to White British patients.
Discussion Variations in-hospital mortality following surgery is important to patients and health service providers. There are a growing number of factors implicated including surgeon caseload, hospital volume and the day of the week operating.22e25 This study found no evidence of detrimental outcomes for any named ethnic group. This would suggest that there are no identifiable barriers (e.g. cultural, linguistic or other) that impact upon the provision of emergency surgical care. The
major caveat to this is the identification of a significant proportion of patients where ethnicity is ‘not stated or known’. This group comprises around 5% of the total cohort, but represents an important, yet un-definable, cohort with significantly poorer outcomes compared to others. The 5% missing data rate for the ethnicity has improved compared to historic estimates of HES;14 this group still represents the second group here. An attempt was made to further reduce missing data by scanning individual patientlevel data for other admissions either before or after the index admission to elucidate ethnicity data that may have been recorded on other occasions. Name recognition software has been used by others but is currently only 7% sensitive26 and individual names are not available on the HES database. The ‘not stated or known’ cohort appeared to have some similarities to certain ethnic groups and differences to others. It is likely to represent a mix of different ethnicities rather than one single ethnic group. This aside it is interesting that the proportion of patients missing ethnicity data varied significantly according to geographical region (Supplementary Fig. 1). The reasons for this are not clear. Unrecorded ethnicity was associated with disproportionately worse in-hospital mortality compared to the rest of the population studied. This is despite a shorter time between
p u b l i c h e a l t h 1 2 9 ( 2 0 1 5 ) 1 4 9 6 e1 5 0 2
admission and an operation in selected emergency surgeries and overall length of stay. One hypothesis for the missing ethnicity data is that the group represents an ‘un-well’ set of patients that require rapid operations. However, an important caveat is this cohort is discharged quickest; this would not be expected if patients had undergone emergency surgery of this type. Another possibility is that these patients come from a setting of enhanced social support and can thereby be discharged earlier, but this would not necessarily account for the earlier operations. There are weaknesses in analysing administrative datasets including continued systematic reporting errors, missed cases and inaccuracies despite improvements in quality assurance.27,28 There is a lack of clinical information including disease severity and general health within the HES dataset such that there are other unknown variables that will not have been fully controlled for. In addition, there are established provider-related biases linked to racial and ethnic prejudice which could not be controlled for here.9 Ethnicity-related variability in both access to surgery and outcomes after surgery identified by others in the elective surgical setting29,30 was not replicated in this cohort of patients for whom ethnicity was recorded. In addition, from the cohort presented here, the variability in the data between providers was low. Taken together this suggests that in the emergency setting the central tenet of the NHS to provide good health care across providers holds true. The group of patients for whom ethnicity is not recorded in part impacts upon the power of this study. The worse outcomes in this group and the increasing ethnic diversity of England mean a strategic endeavour to improve awareness and coding of ethnicity is vital as we appear to be missing and perhaps underserving a high risk group.
Author statements Acknowledgements The authors would like to thank attendees at the Association of Surgeons of Great Britain and Ireland 2014 for their comments.
Ethical approval This study is exempt from UK National Research Ethics Committee approval as it involved analysis of an existing dataset of anonymized data for service evaluation. A data sharing agreement with the Health and Social Care Information Centre (HSCIC) to use Hospital Episode Statistics data has been granted. Studies performed at the Department of Informatics, Queen Elizabeth Hospital Birmingham fall under that arrangement and has been approved by Data Access Advisory Group (DAAG) hosted by the HSCIC.
Funding None.
1501
Competing interests None.
Authors' contributions RSV, TP and FE participated in the conception, design, analysis, writing and editing of this study. All authors participated in the design and writing of the manuscript. FE, IB, DR and PP participated in the statistical analysis. RSV, FE, IB and DR are the guarantors. All authors read and approved the final manuscript.
Appendix A. Supplementary data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.puhe.2015.07.038.
references
1. Isajiw WW. Definition and dimensions of ethnicity: a theoretical framework. In: Challenges of measuring an ethnic World: Science, politics and reality: Proceedings of the Joint CanadaUnited States Conference on the measurement of ethnicity. Washington, DC: U.S. Government Printing Office; April 1e3, 1992. p. 407e27. 2. http://www.ons.gov.uk/ons/rel/census/2011-census/keystatistics-for-local-authorities-in-england-and-wales/index. html [Accessed 01.06.2014]. 3. Levine DA, Neidecker MV, Kiefe CI, Karve S, Williams LS, Allison JJ. Racial/ethnic disparities in access to physician care and medications among US stroke survivors. Neurology 2011;76:53e61. 4. Hassan MO, Arthurs Z, Sohn VY, Steele SR. Race does not impact colorectal cancer treatment or outcomes with equal access. Am J Surg 2009;197:485e90. 5. Fleury AC, Ibeanu OA, Bristow RE. Racial disparities in surgical care for uterine cancer. Gynecol Oncol 2011;121:571e6. 6. Trivedi AN, Grebla RC, Wright SM, Washington DL. Despite improved quality of care in the veterans affairs health system, racial disparity persists for important clinical outcomes. Health Aff (Millwood) 2011;30:707e15. Dao M, Durieux-Paillard S. Quality in 7. Hudelson P, Dominice practice: integrating routine collection of patient language data into hospital practice. Int J Qual Health Care 2013;25(4):437e42. 8. Juckett G. Cross-cultural medicine. Am Fam Physician 2005 Dec 1;72(11):2267e74. 9. Bhogal R. Spectre of racism in health and health care: lessons from history and the United States. BMJ 1998;316:1970. 10. NCEPOD. Emergency admissions: a journey in the right direction? A report of the national confidential enquiry into patient outcome and death. London: NCEPOD; 2007. 11. The Royal College of Surgeons of England. Emergency surgery: standards for unscheduled care. RCSENG e professional standards and regulation, London; 2011. 12. Kluger Y, Ben-Ishay O, Sartelli M, Ansaloni L, Abbas AE, Agresta F, et al. World society of emergency surgery study group initiative on timing of acute care surgery classification (TACS). World J Emerg Surg 2013;8:17.
1502
p u b l i c h e a l t h 1 2 9 ( 2 0 1 5 ) 1 4 9 6 e1 5 0 2
13. Huang Q, Thind A, Dreyer JF, Zaric GF. The impact of delays to admission from the emergency department on inpatient outcomes. BMC Emerg Med 2010;10:16. 14. Ghaferi AA, Birkmeyer JD, Dimick JB. Variation in hospital mortality associated with inpatient surgery. N Engl J Med 2009;361:1368e75. 15. The NHS Information Centre. How good is HES ethnic coding and where do the problems lie? [accessed 28.04.15], http://www. hesonline.nhs.uk/Ease/servlet/ContentServer? siteID¼1937&categoryID¼171; 2011. 16. Jack RH, Linklater KM, Hofman D. Ethnicity coding in a regional cancer registry and in hospital episode Statistics. BMC Public Health 2006;6:281. 17. Saunders CL, Abel GA, El Turabi A, Ahmed F, Lyratzopoulos G. Accuracy of routinely recorded ethnic group information compared with self-reported ethnicity: evidence from the English Cancer patient experience survey. BMJ Open 2013;3. e002882. 18. Burns EM, Bottle A, Aylin P, Darzi A, Nicholls RJ, Faiz O. Variation in reoperation after colorectal surgery in England as an indicator of surgical performance: retrospective analysis of hospital episode statistics. Br Med J 2011;343. d4836. 19. International classification of diseases, 10th revision. World Health Organization; 2010 [Retrieved February 26, 2013]. 20. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987;40:373e83. 21. Carstairs V, Morris R. Deprivation: explaining differences in mortality between Scotland and England and Wales. BMJ 1989;299:886e9. 22. Morris EJA, Taylor EF, Thomas JD, Quirke P, Finan PJ, Coleman MP, et al. Thirty-day postoperative mortality
23.
24.
25.
26.
27.
28.
29.
30.
after colorectal cancer surgery in England. Gut 2011;60(6):806e13. Aylin P, Alexandrescu R, Jen MH, Mayer EK, Bottle A. Day of week of procedure and 30 day mortality for elective surgery: retrospective analysis of hospital episode statistics. BMJ 2013;346. f2424. Bach PB, Cramer LD, Schrag D, Downey RJ, Gelfand SE, Begg CB. The influence of hospital volume on survival after resection for lung cancer. N Engl J Med 2001;345:181e8. Finks JF, Osborne NH, Birkmeyer JD. Trends in hospital volume and operative mortality for high-risk surgery. N Engl J Med 2011;364:2128e37. Ryan R, Vernon S, Lawrence G, Wilson S. Use of name recognition software, census data and multiple imputation to predict missing data on ethnicity: application to cancer registry records. BMC Med Inform Decis Mak 2012;12:3. Sinha S, Peach G, Poloniecki JD, Thompson MM, Holt PJ. Studies using English administrative data (hospital episode statistics) to assess health-care outcomesdsystematic review and recommendations for reporting. Eur J Public Health 2013;23(1):86e92. Westaby S, Archer N, Manning N, Adwani S, Grebenik C, Ormerod O, et al. Comparison of hospital episode statistics and central cardiac audit database in public reporting of congenital heart surgery mortality. BMJ 2007;335(7623):759. Skinner J, Weinstein JN, Sporer SM, Wennberg JE. Racial, ethnic, and geographic disparities in rates of knee arthroplasty among medicare patients. N Engl J Med 2003;349:1350e9. Liu JH, Zingmond DS, McGory ML, SooHoo NF, Ettner SL, Brook RH, et al. Disparities in the utilization of high-volume hospitals for complex surgery. JAMA 2006;296(16):1973e80.