Pooled incidence and case-fatality of acute stroke in Mainland China, Hong Kong, and Macao: A systematic review and meta-analysis.

BACKGROUND: Stroke incidence and case-fatality in Mainland China, Hong Kong, and Macao vary by geographic region and rates often differ across and within regions. This systematic review and meta-analysis (SR) estimated the pooled incidence and short-term case-fatality of acute first ever stroke in mainland China, Hong Kong, and Macao.
METHODS: Longitudinal studies published in English or Chinese after 1990 were searched in PubMed/Medline, EMBASE, CINAHL, Web of Science, SinoMed and CQVIP. The incidence was expressed as Poisson means estimated as the number of events divided by time at risk. Random effect models calculated the pooled incidence and pooled case-fatality. Chi-squared trend tests evaluated change in the estimates over time. When possible, age standardised rates were calculated. Percent of variation across studies that was due to heterogeneity rather than chance was tested using the I2 statistic.The effect of covariates on heterogeneity was investigated using meta-regressions. Publication bias was tested using funnel plots and Egger's tests.
RESULTS: Overall, 72 studies were included. The pooled incidences of total stroke (TS), ischaemic stroke (IS) and haemorrhagic stroke (HS) were 468.9 (95% confidence interval (CI): 163.33-1346.11), 366.79 (95% CI: 129.66-1037.64) and 106.67 (95% CI: 55.96-203.33) per 100,000 person-years, respectively, varied according to the four economic regions (East Coast, Central China, Northeast and Western China) with the lowest rates detected in the East Coast. Increased trends over time in the incidence of TS and IS were observed (p<0.001 in both). One-month and three-to-twelve-month case-fatalities were 0.11 (95% CI: 0.04-0.18) and 0.15 (95% CI: 0.12-0.17), respectively for IS; and 0.36 (95% CI: 0.26-0.45) and 0.25 (95% CI: 0.18-0.32), respectively for HS. One-month case-fatality of IS and HS decreased over time for both (p<0.001). Three-to-twelve-month fatalities following IS increased over time (p<0.001). Publication bias was not found.
CONCLUSIONS: Regional differences in stroke incidence were observed with the highest rates detected in less developed regions. Although 1-month fatality following IS is decreasing, the increased trends in 3-12-month fatality may suggest an inappropriate long-term management following index hospital discharge. REGISTRATION: Registration-URL: https://www.crd.york.ac.uk/prospero/; Reference code: CRD42020170724.

Introduction

Globally, stroke is the second leading cause of death and disability after ischaemic heart disease, accounting for 11.8% of all deaths and is the third most common cause of disability worldwide [1]. Divergent trends in stroke incidence in high- and low-to-middle-income countries have been reported, showing declining and increasing incidences, respectively [2]. Predominantly cross-sectional survey data from China, the largest middle-income country with a rapidly ageing population, suggest a continuous increase in stroke incidence during recent decades [3], with first-ever stroke notifications doubling from 2002 to 2013 [4]. Within China, there are variations in stroke incidence according to a north to south divide, with highest rates in the Northeast and lowest rates in the Southeast regions [5]. Within those regions, there is considerable variation in reporting rates [6]. The pooled estimate of stroke incidence in China is unknown.

The worldwide one-month and up-to-one-year case fatality of ischaemic and haemorrhagic stroke and their temporal trends are divergent [7-9]. Chinese data show a steady decrease in in-hospital mortality following stroke [10], which may reflect improved immediate short-term stroke management over time due to advances in stroke treatment including thrombolysis, and patient care in acute stroke units. Trends in long term (up-to-one year) case fatality by different types of stroke in China are currently unknown. Furthermore, published stroke case-fatalities are inconsistent between and within regions [6], possibly resulting from differences in diagnostic, analytic or design methods, or selected samples that are not representative of the general Chinese population.

China has four economic regions: East Coast, Central China, Northeast and Western China. These regions were formed due to the rapid urbanisation, economic development, and change in Chinese regional population distributions over the past few decades [11]. Of these, the East Coast is the most developed while the Western region is the least developed. Incidence and case fatality of stroke by these regions are unknown. The aim of this novel systematic review and meta-analysis (SR) was to estimate the overall and regional pooled incidence of stroke and one-month and up-to-one-year case-fatality in mainland China, Hong Kong and Macao, using data from longtiduial studies and following strict inclusion and exclusion criteria. Pooled trends of different types of stroke over time were also estimated.

Methods

Search strategy and selection criteria

This SR used the World Health Organization definition of stroke ("rapidly developed clinical signs of focal (or global) disturbance of cerebral function, lasting more than 24 hours or leading to death, with no apparent cause other than of vascular origin") [12]. In addition, the International Statistical Classification of Diseases and Related Health Problems codes I60-I64 (10th version) or equivalent codes in the earlier versions were utilised to identify stroke. Prospective or retrospective cohort studies on adult (aged ≥18 years) populations residing in Mainland China, Hong Kong or Macao, reported in Chinese or English and published after 1990 were eligible for inclusion. When multiple studies used the same study sample, publications with the most complete data were included. Only studies with a sample of more than 100 participants were considered; studies with fewer participants were considered as case series [13]. Studies on recurrent or prevalent stroke were excluded. Studies were also excluded when denominators for estimating the incidence were not extractable.

PubMed/Medline, EMBASE, CINAHL, and Web of Science were searched for English studies while SinoMed and CQVIP were searched for Chinese studies, published from January 1, 1990 to March 13, 2021. The title and abstract were searched for all types of acute stroke and mortality whereas, the "region", and "incidence" were searched in the full text using keywords, subject headings or MeSH terms. English and Chinese peer-reviewed articles were searched using the same strategy (S1 Table). Identified systematic reviews were further hand-searched for other potentially eligible publications. Grey literature, government reports, textbook publications, and unpublished studies were not considered. Overall, 22% of study selection and data extraction was randomly and independently reassessed. Inconsistencies in study selection were discussed in team meetings to reach consensus. If required, authors of identified papers were contacted for additional information before decisions on eligibility were reached.

Registration and protocol

The SR was registered in International Prospective Register of Systematic Reviews [registration code: CRD42020170724]. Details of the study have been reported in the study protocol [14].

Statistical analysis

Study variables included publication details, publication language, study design, type of stroke, diagnostic criteria, follow-up time, study period, regions, setting, sample size, stroke events, stroke severity (National Institutes of Health Stroke Scale and Glasgow Coma Scale), deaths, age, sex, smoking status and inclusion and exclusion criteria of the original study. Incidences of first ever total stroke, ischaemic stroke and haemorrhagic stroke were estimated as events per 100,000 person-years. Using the Exact method, the incidence rates of disease were expressed as Poisson means divided by time at risk. The log incidence rates together with their corresponding log standard errors, stratified by the four economic regions, were meta-analysed using DerSimonian and Laird random effects model. Random effects models were also used to estimate the pooled case fatality estimates. Region-specific analyses of pooled estimates of linear trends over time were assessed using a chi-squared test. The I statistic assessed the heterogeneity of the included studies. Since some studies excluded patients with major comorbidities, subgroup analyses were conducted according to papers that had and did not have this exclusion criterion. Funnel plots and Egger’s tests examined publication bias. Separate analyses were conducted for ischaemic, haemorrhagic, and total stroke. Meta-regressions were constructed to investigate and quantify the proportion of between-study variance explained by known study variables. The meta regression models accounted for age, sex, smoking status, severity of stroke, study setting (community or hospital), follow-up time and risk of bias.

The Newcastle-Ottawa Scale was used to assess the quality and risk of bias. The Healthcare Research and Quality standards were used to further categorise the results of Newcastle-Ottawa Scale assessments into good, fair, or poor quality. Sensitivity analyses were conducted by risk of bias.

Stata/SE 15 was used to conduct the analyses. Stata’s “metan” and “metaprop”commands were used to run the meta-analysis.

Results

A total of 41,294 English, and 46,339 Chinese publications were identified, of which, 32,355 duplicates were removed, leaving 55,278 papers for title and abstract screening. Of these, the full text of 2,440 English and 2,720 Chinese articles was examined, with 72 papers reporting incidence and/or mortality included in the final review with one study reporting both. The reporting of this SR complied with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) (Fig 1).

Fig 1

PRISMA chart, study identification.

Characteristics of the included publications (earliest published in 1997 and latest in 2020) are displayed in Table 1A and 1B. Studies reported on the incidences of total stroke (TS) (n = 28), haemorrhagic stroke (HS) (n = 20) and ischaemic stroke (IS) (n = 20). All except one study followed a prospective cohort design with the exception being retrospective cohort. Of the four economic regions represented in the incidence studies, the east coast was mostly represented (14 studies), followed by western China (five studies), northeast China (four studies), and central China (two studies) and nine were national. In total, the studies reporting incidence included 1,809,252 individuals contributing 13,060,309 person-years of follow-up. Both sexes were represented in most studies with just two studies focusing exclusively on either males or females. Sex was not identified in five incidence studies. The publication years of the 39 studies reporting on case-fatality [IS (n = 28 studies), HS (n = 14 studies)] ranged from 2003 to 2020. Case-fatalities of stroke patients in the East Coast were mostly reported in sixteen studies, followed by Central China (nine studies), Western China (five studies), and Northeast China (four studies). Five studies were national. Sex was not identified in eight case-fatality studies.

Table 1

a. Description of included studies: incidence.

b. Description of included studies: case-fatality.

a
Study	Publication language	Economic region †	Period of study	Risk of bias	Mean age	% Male	Sample size	Type of stroke and number of cases	Follow-up (person-years)	Mean follow-up (years)
Du et al (1997) [15]	Chinese	East Coast	1979–1993	Fair	48.6	50.6	1809	35 TS, 16 HS	25212	NR
Duan et al (2019) [16]	English	East Coast	2010–2016	Fair	49.5	100	23433	678 TS, 595 IS, 109 HS	138255	5.9
Gu et al (2020) [17]	English	East Coast	2009–2014	Poor	69.1	43.8	1257	113 TS	6136	4.8
Guo et al (2013) [18]	Chinese	East Coast	2003–2010	Good	56.2	42.1	10565	195 TS	67886	6.4
Jia et al (2011) [19]	English	East Coast	2006 –NR	Poor	52.9	52.8	19369	341 TS	58107	3
Li et al (2015) [20]	English	East Coast	1992–2012	Poor	NR	NR	3031	908 TS, 664 IS, 231 HS	304260	NR
Liu et al (2014) [21]	English	East Coast	2006–2010	Fair	51.4	79.5	90517	1472 TS, 1049 IS, 423 HS	362163	4.1
Wang et al (2016) [22]	English	East Coast	2002–2012	Good	56	34.8	1263	116 TS, 101 IS	11569	9.2
Wang et al (2018) [23]	English	East Coast	2009–2016	Fair	60.2	40.6	4081	138 IS	20040	5.2
Wang et al (2019) [24]	Chinese	East Coast	2005–2013	Fair	NR	36.4	29689	1716 IS, 91 HS	175667	6.7
Yu et al (2016) [25]	English	East Coast	1996–2011	Fair	51.5	0	64328	2991 TS, 2750 IS, 241 HS	618412	10
Zhang et al (2019) [26]	English	East Coast	2010–2015	Fair	58.4	37.8	8500	444 TS	43350	5.1
Ji et al (2020) [27]	English	East Coast	2009–2017	Poor	45.7	47.8	27712	617 TS	193984	7
Wang et al (2020) [28]	Chinese	East Coast	2005–2013	Poor	52.1	36	28940	2164 IS, 127 HS	177279	6.2
Han et al (2017) [29]	English	Central	2001–2012	Fair	71.7	48.2	2852	211 TS	18628	NR
Zhou et al (2020) [30]	English	Central	2008–2014	Fair	61.7	44.1	31750	1557 TS, 1151 IS, 287 HS	196850	6.2
Li et al (2018) [31]	English	Northeast	2008 –NR	Poor	NR	NR	1928	98 TS	9531	4.9
Sun et al (2013) [32]	English	Northeast	2004–2010	Fair	NR	NR	38949	858 TS, 483 IS, 348 HS	168023	NR
Xie et al (2019) [33]	English	Northeast	2012–2017	Fair	55.6	34.6	3229	81 TS	15499	4.8
Yu et al (2020) [34]	English	Northeast	2004–2014	Fair	50.5	NR	38589	3855 TS	275526	7.1
Guo et al (2016) [35]*	English	Western	2002–2012	Poor	54	52.4	425901	13274 IS, 2917 HS	1895447	NR
Olofindayo et al (2015) [36]	English	Western	2003–2012	Fair	46.5	40.9	2589	121 TS, 75 IS, 44 HS	23292	9.2
Ren et al (2019) [37]	Chinese	Western	2011–2015	Poor	46.8	61	33042	448 TS	80506	2.4
Zhang et al (2009) [38]	English	Western	2002–2008	Fair	64.3	45.2	2173	82 TS, 52 IS, 30 HS	9952	4.6
Zhou et al (2016) [39]	Chinese	Western	2011–2014	Poor	NR	NR	4160	51 TS	13621	3.3
Huang et al (2003) [40]	Chinese	National	1994–2001	Poor	54.7	51.8	24475	228 TS, 151 IS, 77 HS	105201	4.3
Huang et al (2019) [41]	English	National	1992–2015	Good	50.9	41	117575	2276 IS, 1019 HS	900214	NR
Kelly et al (2008) [42]	English	National	1991–2000	Good	55.9	49.1	155131	2421 HS	1287587	8.3
Liu et al (2008) [43]	Chinese	National	1992–2003	Poor	46.9	53.5	30384	613 TS, 427 IS, 159 HS	199711	NR
Wang et al (2013) [44]	English	National	1987–1998	Fair	51.6	47.2	26607	1108 TS, 614 IS, 451 HS	241149	NR
Zhang et al (2007) [45]	Chinese	National	1992 –NR	Poor	NR	46.4	15131	370 TS, 266 IS, 107 HS	163415	10.8
Zhou et al (2003) [46]	Chinese	National	1991–1997	Fair	45	48.6	9111	259 TS, 177 IS, 84 HS	131188	NR
Chen et al (2020) [47]	English	National	2004–2017	Fair	51.6	40.9	489586	45732 TS, 36588 IS, 8142 HS	4720000	NR
Swaminathan et al (2020) [48]	English	National	2003–2019	Fair	50.5	41.9	41596	1752 TS	402649	9.7
b
Study	Publication language	Economic region †	Type of stroke	Period of study	Risk of bias	Mean age	% Male	Cases	Deaths	Follow-up
Liu et al (2018) [49]	Chinese	East Coast	IS	2013–2017	poor	68	54.5	101	14	3 months
Shi et al (2019) [50]	English	East Coast	IS	2010–2017	good	NR	NR	334	28	3 months
Geng et al (2016) [51]	Chinese	East Coast	IS	2013–2015	fair	NR	NR	796	37	3 months
Fang et al (2012) [52]	English	East Coast	IS	2003–2008	good	66.7	56.3	1184	113	28 days
Tu et al (2017) [53]	English	East Coast	IS	2015	fair	58	54.3	737	104	3 months
Zi et al (2013) [54]	English	East Coast	IS	2009–2012	fair	65	65.9	226	34	3 months
Chen et al (2016) [55]*	English	East Coast	IS	2013–2015	good	68.4	58.4	871	94	3 months
Huang et al (2013) [56]	English	East Coast	IS	2008–2011	fair	66.3	58	338	68	12 months
Liu et al (2020) [57]*	Chinese	East Coast	IS	2012–2019	poor	72	50.4	117	20	3 months
Xi et al (2020) [58]	Chinese	East Coast	IS	2016–2019	poor	73	54.3	138	16	3 months
Hong et al (2016) [59]	Chinese	East Coast	IS	not given	fair	NR	NR	2479	255	3 months
								2479	349	12 months
Wang et al (2018) [60]*	English	East Coast	HS	2016–2017	fair	65.8	61.9	181	35	1 month
Wei et al (2014) [61]	English	East Coast	HS	2010–2012	fair	69	53.1	271	34	3 months
Yan et al (2016) [62]	English	East Coast	HS	2011–2014	poor	63.2	58.9	112	33	6 months
Jiang et al (2018) [63]	Chinese	East Coast	HS	2016–2017	fair	63	66.3	172	12	30 days
Hu et al (2012) [64]	English	East Coast	HS	2011	fair	63.1	61.4	176	64	3 months
Nie et al (2017) [65]	English	Central	IS	2013–2016	fair	59	53	387	74	12 months
Wang et al (2014) [66]	English	Central	IS	2011–2014	fair	68	50.9	275	75	12 months
Zhu et al (2018) [67]	English	Central	IS	2016–2017	fair	66	48.2	299	31	3 months
Jing et al (2019) [68]	English	Central	IS	2016–2018	fair	65	58.2	196	43	12 months
Sun et al (2015) [69]	English	Central	IS	2004–2010	good	NR	NR	790	73	30 days
								790	116	12 months
Lin et al (2014) [70]	Chinese	Central	HS	2010–2012	fair	58.2	62.9	167	58	6 months
Zhang et al (2014) [71]*	English	Central	HS	2007–2013	poor	72.6	59.3	118	50	1 month
Yang et al (2004) [72]	English	Central	HS	not given	poor	69.7	NR	722	407	28 days
Jiang et al (2011) [73]	Chinese	Central	HS	2009–2010	fair	61.5	55.5	182	57	30 days
Wang et al (2014) [74]	English	Northeast	IS	2012–2013	fair	65	60.4	326	38	1 month
Dong et al (2013) [75]	English	Northeast	IS	2010–2011	poor	69	55.2	125	18	3 months
Zhang et al (2013) [76]	English	Northeast	IS	2007–2011	fair	72	58	245	41	12 months
Xie et al (2016) [77]	English	Northeast	IS	2013–2014	fair	61	64.8	216	32	6 months
Du et al (2018) [78]	Chinese	Western	IS	2013–2017	poor	70.8	55.7	122	34	3 months
Zhang et al (2007) [79]*	English	Western	IS	2000–2006	fair	NR	NR	1904	327	1 month
Tao et al (2017) [80]	English	Western	HS	2010–2013	fair	58.5	64.3	336	90	3 months
He et al (2018) [81]	English	Western	HS	2015–2016	fair	NR	NR	221	47	3 months
Li et al (2008) [82]	English	Western	IS	not given	fair	65.5	57	395	32	3 months
			HS	not given		65.5	57	254	39	3 months
Wu et al (2014) [83]	English	Nation wide	IS	2007–2010	fair	63	64.9	1229	144	12 months
Tu et al (2017) [84]	English	Nation wide	IS	2012–2015	fair	68	54.4	4215	906	12 months
Tu et al (2019) [85]	English	Nation wide	IS	2015–2017	fair	65	52	1530	325	6 months
Zhang et al (2003) [86]	English	Nation wide	IS	1991–2000	good	NR	NR	6772	1144	28 days
			HS	1991–2000		NR	NR	3489	1691	28 days
Chen et al (2020) [47]	English	Nation wide	IS	2004–2017	fair	59.3	44.8	36588	1098	28 days
						59.3	44.8	36588	2267	12 months
			HS	2004–2017	fair	59.3	51.3	8142	3630	28 days

Notes

*Studies marked with asterisk followed a retrospective design, while others were prospective cohort.

†Four economic regions: East Coast (Beijing, Tianjin, Hebei, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Guangdong and Hainan); Central (Shanxi, Anhui, Jiangxi, Henan, Hubei and Hunan); Northeast (Liaoning, Jilin and Heilongjiang); Western (Inner Mongolia, Guangxi, Chongqing, Sichuan, Guizhou, Yunnan, Tibet, Shaanxi, Gansu, Qinghai, Ningxia and Xinjiang)

Abbreviation: Not reported (NR)

Of the 34 studies reporting incidence, 22 (65%) were classified as having good to moderate quality while 12 (35%) were deemed poor quality. The 39 papers reporting case-fatality included 31 (79%) with good to fair quality while 8 (21%) were rated as poor. The quality assessment details can be found in S2 Table. The percentage of agreement among co-authors (FH, HX, TR, GM) and collaborators (YZ, YH, ZP, SL, RY) conducting the culling and quality assessments ranged from 88% to 95%, with the corresponding Cohen Kappa values ranging from 0.48 to 0.56. The overall percentage of agreement was 92%, Cohen Kappa 0.51 [95% confidence interval (CI): 0.45–0.58].

Geographical differences in the pooled incidences of total stroke were found with the highest being in Central China (887.87 per 100,000 person-years) and the lowest in the East Coast region (409.34 per 100,000 person-years). The overall incidence was 468.90 (95% CI: 163.33–1346.11) per 100,000 person-years (Fig 2), increasing to 541.01(95% CI: 149.51–1957.68) per 100, 000 person-years when poor quality studies were excluded from the analysis. The combined pooled estimate in Central, Northeast and Western China of 698.52 (95% CI: 74.87–6517.18) per 100,000 person-years was significantly higher than that in the East Coast region (409.34 (95% CI: 73.22–2288.38)), p<0.001 (S1 Fig).

Fig 2

Total stroke incidence rate per 100,000 person-years.

Similar regional differences were detected for both ischaemic and haemorrhagic stroke when comparing overall pooled estimates for Central, Northeast and Western China with the East Coast region. For IS the pooled estimate of 505.26 (95% CI: 49.47–5160.46) compared to 438.33 (95% CI: 66.01–2910.54) per 100,000 person-years for the East Coast Region while, for HS, the pooled estimate of 165.08 (95% CI: 31.91–854.01) compared to 57.41 (95% CI: 18.24–180.65) per 100,000 person-years for the East Coast region (S2 and S3 Figs). No between-studies heterogeneity was detected in all types of stroke and no publication bias was found as shown in S4 Fig. Egger’s tests p values for TS, IS and HS were p = 0.17, p = 0.39, and p = 0.43, respectively. The regional differences observed in the sensitivity analysis were consistent with our main findings (S5 Fig). Age standardisation was limited to a few selected studies based on data availability resulting in an estimated age standardised incidence for TS, IS, and HS of 285.76 (95% CI: 29.37–2780.33), 586.76 (95% CI: 7.05–48813.46) and 141.55 (95% CI: 4.58–4379.29) per 100, 000 person-years, respectively.

Fig 3 shows the scatter plots for stroke incidence using the median study period year for each study. Studies published in a period of four consecutive years were pooled together, starting from the latest year. The analyses showed a statistically significant increase over time in total stroke, ischaemic stroke and haemorrhagic stroke (Chi-squared test p value<0.001 in all).

Fig 3

Scatter plots of stroke incidence by median study period.

Subgroup analyses were based on initial study sample selection criteria. Studies that included a more representative sample (without the exclusion of patients with major comorbidities such as coronary heart disease) were categorised as group one. Studies that used selected samples, often only including healthier individuals, which made the samples less representative of the general Chinese population, were categorised as group two. Pooled incidence of TS in the more representative sample was slightly higher than that found in group two, 474.09 (95% CI: 38.47–5842.50) per 100,000 person-years versus 467.79 (95% CI: 146.36–1495.20) (S6 Fig).

The pooled estimates of one-month and three-to-twelve-month case-fatality were 0.11 (95% CI: 0.04–0.18) and 0.15 (95% CI: 0.12–0.17), respectively for IS; and 0.36 (95% CI: 0.26–0.45) and 0.25 (95% CI: 0.18–0.32), respectively for HS (Figs 4 and 5). Excluding studies with poor quality produced similar findings. High between-study heterogeneity was found when studies were stratified by follow-up time (I >90%). Based on meta-regressions, risk of bias, study setting (community or hospital), and follow-up time explained between 11.4% to 17.4% of between-study variance in IS. For HS, risk of bias, study setting, and smoking status explained between 51.7% to 73.5% of between-study variance. Mean age, sex, and severity of stroke did not contribute to study heterogeneity in either type of stroke.

Fig 4

Pooled case-fatalities of ischaemic stroke.

Fig 5

Pooled case-fatalities of haemorrhagic stroke.

Following the same method used in subgroup analysis of stroke incidence, one-month estimates of IS were similar in both groups. However, these varied for HS with pooled estimates being higher in the more representative sample [0.49 (95% CI: 0.44–0.55) versus 0.25 (95% CI: 0.10–0.40)] in the less representative group.

The scatter plots of case-fatality of IS and HS using the median study period year for each study are shown in Fig 6. Decreased trends in 1-month IS and HS case-fatality were detected (p<0.001 in both), whereas, 3-12-month IS case-fatality increased over time (p<0.001). Trends in 3-12-month HS case fatality could not be investigated due to limited points in time (n = 3) after studies were pooled by median years.

Fig 6

Scatter plots of stroke case-fatality by median study period.

Discussion

This is the first study to report pooled estimates of incidence and case-fatality of stroke in China using data from longitudinal cohort studies. Our estimated figures support previous findings that China has one of the highest rates of stroke incidence in the world [87]. We report significant differences in incidences of TS, IS and HS between the East Coast (with lowest rates) and other economic regions (with higher rates) in China. This SR reports that although 1-month fatality following ischaemic stroke has decreased over the past decades, 3-12-month mortality following the event has significantly increased during this time.

Rates of stroke incidence rates are often reported inconsistently in China [6]. Such inconsistensies in reportings may have resulted from different diagnostic and definitional criteria [6], and different study designs including door-to-door surveys that lack follow-up periods and potentially generate biased estimates due to sampling errors, non-response and reporting bias [88]. As detected in our SR, studies reporting stroke incidence have often excluded patients with histories of certain comorbidities [27], making the sample less representative of the general Chinese population at risk of stroke. Regional differences in stroke incidence may also be explained by the different patterns of risk factors reported across China. For instance, the higher prevalence of hypertension in Northern, Northeast and Central China may have contributed to a higher incidence of stroke [3]. The disparities in stroke incidence among the four economic regions, identified in this SR, highlight the need for primary stroke prevention, health promotion, and better access to health services in less-developed regions.

Consistent with previous research [4], this SR reports an increased incidence of TS, IS and HS over time, which is consistent with findings from other low to middle countries that are experiencing a similar trend [2]. As a result of rapid economic development, there is evidence to indicate that populations living in different parts of China have changed their lifestyle, their diet (consuming more fast foods), and have become less physically active [89]. This may have led to higher rates of obesity, hypertension and diabetes, which in turn can lead to higher rates of stroke [90]. The increase in incidence may also be explained by the rapidly aging population or improvements in stroke diagnosis [4, 91]. In contrast, stroke incidence in high income countries significantly declined from 1990 to 2010 [2]. This may indicate that China lags behind in stroke prevention despite the rapid economic growth experienced in the same period.

This SR reports decreasing trends in one-month IS and HS case-fatality, consistent with a previous study that reported decreased in-hospital mortality for all types of stroke [10]. Underlying factors contributing to the downward trend in short-term stroke case-fatality in China may include improved treatments, expanded healthcare coverage, enhanced preventative campaigns and increased public health literacy around stroke [92]. This study, however, also found an opposing increase in 3-12-month IS case-fatality, which may indicate a need to improve secondary prevention, post-stroke rehabilitation and support for caregivers, especially for those discharged from acute care. This is supported by a study that found that Chinese stroke patients were less likely to adhere to antiplatelet and lipid lowering therapy after 3 months post discharge, with a large proportion of these patients not filling prescriptions or refusing to take prescribed medications [93]. Barriers for patients to adhere to post-stroke medication regimens include limited health literacy, inadequate communication and poor coordination between medical staff and patients, and cost [94]. Although recommended by guidelines, a national study reported that around 40% of IS patients were not properly rehabilitated with rehabilitation services varying across hospitals [95]. Underlying reasons for such suboptimal stroke rehabilitation may include lack of insurance coverage, absence of well-established rehabilitation systems, a lack of trained and licensed therapists and caregivers, and limited availability of rehabilitation technologies [96]. Furthermore, due to the barriers to accessing government-funded services, post-stroke patient care often falls on family members who lack medical skills and have little social support to properly manage discharged patients at home [97, 98].

This SR highlights the need to tackle inequality in stroke prevention across the four economic regions by reinforcing and concretising policies. This SR also emphasises the need to establish well-coordinated rehabilitation system programs that will meet the needs of individuals coming from different socioeconomic backgrounds.

Strengths and limitations

The comprehensive search strategy targeted six major electronic databases to identify potentially eligible English and Chinese articles. The earliest publication year was restricted to 1990 since brain imaging technologies were widely used after this year in China [99], which made the case ascertainment of stroke more reliable. The restriction to 1991 onwards aimed to decrease false positives and to increase the validity of included cases. Our strict selection criteria, that included longitudinal studies with a defined follow-up period, ensured the inclusion of only acute first-ever stroke. This study, covering a time period of four decades (1979 to 2019), included around 2 million participants of all ages and over 13 million person-years of follow-up.

This SR also has limitations. Less developed regions were under-represented which may indicate that the regional differences we found could, in reality, be greater. The age-standardised rates were mostly based on studies from the East Coast, which had lower estimates than other regions. This may explain the large differences in the crude and age-standardised rates. Duplicate research populations were carefully screened for using all available information with papers reporting on the same study population included once. There may, however, have been some overlap in study populations from the same region in studies which included limited descriptions of participants. Subtypes of IS and HS were not evaluated in this SR. The sources of heterogeneity investigated in our study were limited to available variables reported in the included studies.

Conclusions

The disparities in stroke incidence between economically developed regions and less developed regions necessinate decision-making and action to address inequalities in stroke education and prevention across China. The rising trends in 3-12-month case-fatality following ischemic stroke highlight the need to develop more effective secondary stroke prevention programs, including higher adherence to antiplatelet and lipid lowering therapies and improvement of post stroke rehabilitation.

PRISMA checklist.

(DOCX)

Click here for additional data file.

Search strategy.

(DOCX)

Click here for additional data file.

Risk of bias assessments.

(DOCX)

Click here for additional data file.

Pooled estimates of total stroke incidence in east coast China and other regions.

(TIFF)

Click here for additional data file.

Pooled estimates of ischaemic stroke incidence in east coast China and other regions.

(TIFF)

Click here for additional data file.

Pooled estimates of haemorrhagic stroke incidence in east coast China and other regions.

(TIFF)

Click here for additional data file.

Funnel plots and Egger’s tests.

(TIFF)

Click here for additional data file.

Sensitivity analysis for total stroke incidence.

(TIFF)

Click here for additional data file.

Total stroke incidence rate by sub-group analysis.

(TIFF)

Click here for additional data file.

Pooled estimates of total stroke incidence in map of China.

(TIFF)

Click here for additional data file.

Total stroke incidence by length of follow up.

(TIFF)

Click here for additional data file.

Pooled estimates of ischaemic stroke incidence by the four Chinese economical regions.

(TIFF)

Click here for additional data file.

Pooled estimates of haemorrhagic stroke incidence by the four Chinese economical regions.

(TIFF)

Click here for additional data file.

Included Chinese papers and translations.

(DOCX)

Click here for additional data file.

27 Sep 2021

19 Oct 2021

20th October 2021

Dear Prof. Miguel A. Barboza,

Re: PONE-D-21-22996 – “Pooled incidence and case-fatality of acute stroke in China: a systematic review and meta-analysis”

Thank you for the opportunity to revise the manuscript which the reviewer described as “well structured” and “comprehensive” highlighting the importance of our research project as being “very well discussed”. In line with the editor’s and reviewer’s comments, as requested, we have responded on a point-by-point basis and have changed the manuscript accordingly.

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

Response: The manuscript has been modified against these style requirements.

2. Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript. If you need to cite a retracted article, indicate the article’s retracted status in the References list and also include a citation and full reference for the retraction notice.

Response: As suggested, the reference list has been checked ensuring it is complete and correct. Retracted papers have not been cited in the manuscript.

The reference list has been revised to support the added points in the discussion as suggested by the reviewer. The reference of the accepted protocol of this study has also been added.

Response to reviewers’ comments

Reviewer #1:

Comment 1:

The systematic review and meta-analysis is very well structured, ideas are clear and the writing is concise. I congratulate the author/authors for this. The review of literature is comprehensive and the importance of the research is very well discussed as far as applied perspective and theoretical aspect is concerned. The author/ authors have done good reference to the theoretical fundament for the article.

The tables, figures are helpful to understand the content of the paper.

Response: We thank the Reviewer for these positive comments.

Comment 2:

1. The article mentions that there is varying incidence of acute stroke in the the four economic regions with the lowest rates detected in the east coast. This could be because of better healthcare facilities and affordability of population in that region. Increased trends over time in the incidence of TS and IS were observed possibly because of major changes in lifestyle over a period of 4 decades. A three-to-twelve-month fatality following IS increased over time. This could be because of poor rehabilitation, affordability, inadequate follow up and non-compliance with the treatment. This could be because of poor counselling and information to the caregivers about the post stroke care of the patient.

Response: We thank the Reviewer for these insightful comments. Regional differences in stroke incidence in China have been continuously reported, often explained in the literature as a reflection of different prevalence of major risk factors, including hypertension, smoking, drinking and unhealthy diet. The unbalanced economic development in different regions may have impacted the health care system across different regions. Although populations residing in more developed regions (such as the East coast) may be utilizing health care services more than those residing in less developed regions. Following a comprehensive literature search we could not find any literature to support the notion that inequality in health resources distribution may have contributed to regional differences in stroke incidence. Therefore, we prefer to be more conservative when discussing these differences by only emphasizing the different patterns of risk factors across these regions.

As suggested, the possible association between increasing stroke incidence and changing lifestyles is now discussed in the revised manuscript. As mentioned in the original manuscript, improving post-stroke rehabilitation may decrease the three-to-twelve-month fatality following ischaemic stroke (IS). In the revised manuscript we further discussed post stroke rehabilitation and its importance to study outcomes (please see page 19). Affordability is indeed a problem in stroke care in China, which may affect patients’ adherence to secondary stroke prevention medications and access to rehabilitation. The issue of affordability has also been added to this revised manuscript. Non-compliance reflected by nonadherence to medication and insufficient support for the caregivers have similarly been discussed.

It is possible that inadequate follow up could have contributed to increasing three-to-twelve-month fatality following IS but we could not support this statement by published literature and therefore we did not add this point to the revised discussion.

Comment 3

2. This meta-analysis will help in:

a. Reinforcement and concretisation of policies by the policymakers in China to plan according to social and cultural variations of their four economic regions.

b. Planning strategies for primordial, primary and secondary prevention and allocating funds accordingly for infrastructural development.

c. Employing trained workforce in the field of neurosciences for stroke population. Also, developing multi-task force which will comprise of trained healthcare workers, community leaders, private and government sectors and other non-governmental organisations in China.

Response: We thank the Reviewer for these insightful points, which we added to the revised discussion.

Yours sincerely,

Fan He

La Trobe University, Melbourne

Submitted filename: Response to reviewers_v6.docx

Click here for additional data file.

25 Nov 2021

24 Dec 2021

24th December 2021

Dear Prof. Miguel A. Barboza,

Re: PONE-D-21-22996R1 – “Pooled incidence and case-fatality of acute stroke in Mainland China, Hong Kong, and Macao: a systematic review and meta-analysis”

Thank you for the opportunity to revise the manuscript. In line with the editor’s and reviewer’s comments, as requested, we have responded on a point-by-point basis and have changed the manuscript accordingly.

Journal Requirements:

1. Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript. If you need to cite a retracted article, indicate the article’s retracted status in the References list and also include a citation and full reference for the retraction notice.

Response: As suggested, the reference list has been checked ensuring it is complete and correct. Retracted papers have not been cited in the manuscript.

We slightly revised the title of this manuscript making it consistent with the protocol paper that is currently in print.

• He F, Blackberry I, Yao L, Xie H, Mnatzaganian G. Geographic Disparities in Pooled Stroke Incidence and Case Fatality in Mainland China, Hong Kong and Macao: Protocol for a Systematic Review and Meta-Analysis. JMIR Res Protoc. Forthcoming. doi:10.2196/32566

Response to reviewers’ comments

Comment 1:

If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #2: (No Response)

No Response

Comment 2:

2. Is the manuscript technically sound, and do the data support the conclusions? The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #2: Partly

Response: We have responded to this comment. Please find our reasoning and responses detailed in the subsequent reviewer’s comments below

Comment 3

Has the statistical analysis been performed appropriately and rigorously?

Reviewer #2: Yes

Response: We thank the Reviewer for this positive comment.

Comment 4

Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #2: Yes

Response: We thank the Reviewer for this positive comment.

Comment 5

Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #2: No

Response: This manuscript has been edited by a native English speaker.

Comment 6

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #2: I will focus on methods and reporting

No Response

Major

1) funnel plots and publication bias tests only make sense in the presence of an intervention, they are meaningless for incidence rates and in this context.

Response: We respectfully disagree with the reviewer.

We used funnel plots together with Egger’s tests to investigate whether the studies included in the systematic review had a biased representation of higher or lower incidences; that is, publication bias.

One criticism of meta-analyses is that often the available studies may not be representative of all studies addressing the research question. Publication bias refers to the claim that studies with statistically significant results are more likely to be published than studies without. Publication bias can occur in other instances besides trial study designs (Copper et al 2009). An example of a possible scenario is when a hospital director opposes the publication of results that show very high fatality rates. The funnel plot and Egger’s tests investigated whether our findings on between study heterogeneity was due to publication bias.

• Cooper H et al. The handbook of research synthesis and meta-analysis. 2nd edition. Russel Sage Foundation. New York. 2009

Running funnel plots with Egger’s test or other tests that investigate asymmetry is very commonly done in systematic reviews in observational incidence studies. We refer the reviewer to the following publications, which are a few of many that followed this very acceptable statistical method.

• Megan R et al. Systematic review and meta-analysis of the incidence rate of Takayasu arthritis. Rheumatology, 2020; 60(11):4982-4990.

• Makin SDJ, et al. Cognitive impairment after lacunar stroke: systematic review and meta-analysis of incidence, prevalence and comparison with other stroke subtypes. J Neurol Neurosurg Psychiatry. 2013; 84:893-900.

• Hackett ML, et al. Part I: frequency of depression after stroke: an updated systematic review and meta-analysis of observational studies. Int J Stroke. 2014, 9:1017-25.

• Li M et al. Hyperuricemia and risk of stroke: A systematic review and meta-analysis of prospective studies. Atherosclerosis, 2014:265-270.

• Taylor M et al. Longevity of complete dentures: a systematic review and meta-analysis. J Prosthetic Dentistry. 2021; 125:611-617

• Fasugba O et al. Ciprofloxacin resistance in community- and hospital acquired Escherichia coli urinary tract infections: a systematic review and meta-analysis of observational studies. BMC Infect Dis. 2015, 15:545.

• Qin B, et al. Epidemiology of primary Sjögren's syndrome: a systematic review and meta-analysis. Annals of the Rheumatic Diseases 2015;74:1983-1989.

• Meng K, et al. "Incidence of venous thromboembolism during pregnancy and the puerperium: a systematic review and meta-analysis." The Journal of Maternal-Fetal & Neonatal Medicine 28.3 (2015): 245-253.

2) Language corrections are needed.

Response: This manuscript has been edited by a native English speaker.

3) Some justification is needed on the 1990 time limit.

Response: The decision to exclude studies published before 1991 was made to better capture studies that followed similar diagnostic criteria. Before 1991, clinicians would have diagnosed stroke without brain imaging technologies which could have resulted in the inclusion of false positive cases. The restriction to 1991 onwards aimed to decrease false positives and to increase the validity of included cases. We have added a clarification in the revised manuscript.

4) Meta-regression is a stab in the dark usually and is underpowered to detect anything but massive associations (effectively a regression with X observations, where X is the number of available studies). You should discuss this as a major limitation. Even with 60 or 80 studies, it can provide little insight.

Response: Meta regression offers a systematic approach to explain sources of wide variation in reported findings. In our incidence studies, no heterogeneity was detected and, therefore, there was no need to conduct further analyses. However, in the case-fatality analyses, we conducted meta regression to explain between study variances based on the available study variables.

Meta-regression is an extension to subgroup analyses that allows the effect of continuous, as well as categorical characteristics to be investigated, and in principle allows the effects of multiple factors to be investigated simultaneously (although this is rarely possible due to inadequate numbers of studies) (Thompson 2002). According to scientific literature, meta-regression should generally not be considered when there are fewer than 10 to 15 studies in a meta-analysis. Our case-fatality analysis included 39 studies.

1. Cochrane handbook for systematic reviews of interventions. https://training.cochrane.org/handbook/current

2. Higgins JPT, et al. Measuring inconsistency in meta-analysis. BMJ Clinical Research. 327:557-60.

3. Thompson SG, et al. How should meta-regression analyses be undertaken and interpreted? Statistics in Medicine. 2002; 21:1559-1573

4. https://handbook-5-1.cochrane.org/chapter_9/9_6_4_meta_regression.htm

We acknowledge that exploring all sources of heterogeneity is limited as there are many unknown variables that cannot be accounted for. This is always true in all systematic reviews (Copper et al, 2009). Furthermore, the limited number of included studies in the meta-regression add uncertainty in the true sources of heterogeneity. These limitations have been added to the revised manuscript.

5) Report the confidence intervals for I^2 (calculated using heterogi or metaan in Stata) as argued in http://www.ncbi.nlm.nih.gov/pubmed/17974687. A simple formula exists in the seminal 2002 Higgins paper that proposed I^2.

Response: The heterogeneity in case-fatality analyses was very high and statistically significant (p<0.001). As requested, using Borenstein and colleagues methods, we have estimated confidence intervals for the statistic I-squared..

Borenstein M, et al. Introduction to meta-analysis. John Wiley & Sons, Ltd. UK. 2011

Minor

1) Add some more info on the methods section in the abstract, was heterogeneity assessed, publication bias assessed (if relevant) etc.

Response: As suggested, we have revised the abstract to include this.

2) Clarify that the estimates were back transformed for the forest plots

Response: The methods we used are stated in the manuscript: “Using the Exact method, the incidence rates of disease were expressed as Poisson means divided by time at risk. The log incidence rates together with their corresponding log standard errors, stratified by the four economic regions, were meta-analysed using DerSimonian and Laird random effects model. Random effects models were also used to estimate the pooled case fatality estimates”.

3) The axis titles on the forest plots are not clear, reported as percentages when they are supposed to be rates. Currently the methods section and the graphs are not in agreement.

Response: As shown in Figure 2, the title of axis is incidence rate. While, in Figures 4 and 5, the title of axis is proportion.

4) How was the random-effect model implemented, i.e. how was heterogeneity estimated? There are numerous ways to do so. Did they use the standard DerSimonian-Laird method? If so, please state so. Also there are better performing methods, for example please see https://www.ncbi.nlm.nih.gov/pubmed/28815652

Response: This has been stated in the section of “Statistical analysis”: “The log incidence rates together with their corresponding log standard errors, stratified by the four economic regions, were meta-analysed using DerSimonian and Laird random effects model. Random effects models were also used to estimate the pooled case fatality estimates.” We have revised the forest plots showing the case fatalities – now including the weights that were generated by the random effect model.

5) What Stata commands were used? Again this relates to the point about the methods and the graphs.

Response: We used metan and metaprop Stata commands for incidence and case fatality studies, respectively. We have added this clarification in Methods.

6) Clarify the variables included in the meta regression in the methods section.

Response: The meta-regression models included risk of bias, study setting (community or hospital), follow-up time, smoking status, age, sex, and severity of stroke. This has been clarified in the methods section.

Yours sincerely,

Fan He

La Trobe University, Melbourne

Submitted filename: Response to reviewer 2_v2_final.docx

Click here for additional data file.

14 Jun 2022

Pooled incidence and case-fatality of acute stroke in Mainland China, Hong Kong, and Macao: a systematic review and meta-analysis

PONE-D-21-22996R2

Dear Dr. He,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Salvatore De Rosa

Academic Editor

PLOS ONE

17 Jun 2022

PONE-D-21-22996R2

Pooled incidence and case-fatality of acute stroke in Mainland China, Hong Kong, and Macao: a systematic review and meta-analysis

Dear Dr. He:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Salvatore De Rosa

Academic Editor

PLOS ONE