Serotype distribution of Streptococcus pneumoniae and potential impact of pneumococcal conjugate vaccines in China: A systematic review and meta-analysis

ABSTRACT

Objective: Thirteen-valent pneumococcal conjugate vaccines (PCV13) was licensed for optional use in mainland China since 2017, but the uptake is low. To update the research evidence for the pneumococcal serotype distribution of pre-PCV era and to estimate the potential impact of PCVs, we performed a meta-analysis on the relevant publications concerning the Chinese population.

Methods: This systematic review and meta-analysis were conducted on the pneumococcal serotype distribution publications in mainland China from 2000 to 2016. The literature was searched in PubMed, Ovid-EMBASE, Web of Science, CNKI and Wanfang. Heterogeneity and publication bias were tested by I², meta-regression, Egger's and Begg's test. The pneumococcal serotype and vaccine serotype coverage rates were pooled using the random-effects model in Stata SE 12.0.

Results: In total, 85 publications were included. Of all 16,945 included pneumococcal isolates, the most common serotypes/serogroups were 19F, 19A, 23F, 14, and 6B, that from children were the same as above, that from adults≥18 years were 19, 3, 6, 23, and 14. Among isolates from children <18 years, the pooled coverage for PCV10 serotypes was 52.3%, that for PCV13 was 68.4% and that for PPSV23 was 65.5%. Regarding individuals ≥18 years, the pooled coverage for PCV10 serotypes was 29.7%, that for PCV13 was 49.5% and that for PPSV23 was 50.7%. Serotype prevalence and vaccine serotype coverage varied by age group, source, and region.

Conclusions: The most common pneumococcal serotype in mainland China was 19F. The serotype coverage rates of PCV13 and PPSV23 were 50%–68% in mainland China.

Introduction

Streptococcus pneumoniae is an important cause of community acquired infections and is associated with significant morbidity and mortality worldwide. By the end of the 20th century, pneumonia was the first and second leading cause of death among children aged under 5 years in the rural and urban areas of China, respectively.¹ S. pneumoniae causes at least 18% of severe episodes and 33% of deaths from pneumonia worldwide.² The burden of invasive pneumococcal diseases (IPD) has decreased substantially due to the widespread usage of pneumococcal conjugate vaccines (PCVs) worldwide.^3,4 However, S. pneumoniae still poses a significant burden on individuals and healthcare systems because of the high costs of PCVs and the serotype shift after immunization.⁵

S. pneumoniae has more than 90 serotypes, which vary by age and different regions. Only when the distribution of S. pneumoniae and the serotype coverage of PCVs or pneumococcal polysaccharide vaccine (PPSV) are well-known can useful information be offered to guide the vaccination program. In mainland China, 7-valent pneumococcal conjugate vaccine (PCV7) was licensed as type II self-paid vaccine for optional use in 2008 but the extent of adoption was very low. PCV13 has been licensed for optional use since November 2016. However, it won't be widely used due to low awareness and high price. In 2013, WHO provided a new plan that the completed vaccination coverage rate of PCVs should reach 90% to avoid severe pneumonia in children by 2025.⁶ Before the use of PCV13 around China, it is important to update the research evidence for pneumococcal disease. Currently, many papers have reported on pneumococcal serotypes in mainland China, but there is a lack of evidence of the pneumococcal serotype distribution countrywide and in all age groups. A recently published manuscript has reviewed the pneumococcal serotype distribution and coverage among Chinese children, but they did not include isolates from adults or provide detailed information regarding the serotype distribution and vaccine serotype coverage by age group.⁷ Thus, to provide a comprehensive view of the pneumococcal isolate characteristics and estimate the potential impact of vaccines in China, we performed a systematic review and meta-analysis of publications on serotype distributions of pneumococcal infections in mainland China from 2000 to 2016.

Results

Literature characteristics

In total, 1,848 potentially related articles were identified, and 384 articles were duplicated. Of these articles, 123 were identified as potentially relevant articles according to the screening of the titles and abstracts. Six articles were excluded because they were reviews, 16 articles shared the same data, and 4 articles had a sample size that was too small. Finally, 85 articles were included in our meta-analysis (Fig. 1). The characteristics of the included articles were listed in Table 1. The regional distribution of the enrolled articles and pneumococcal isolates are shown in Fig. 2.

Figure 1.

Selection of Articles for Meta-analysis.

Table 1.

Characteristics of the included studies published from 2000 to 2016 in mainland China.

								Vaccine Coverage Rate a
	First Author	Study Year	Age (y or m)	Location in China	Source	Number of strains	Serotype method	PCV10	PCV13	PCV23
South
	Yang F20	1996–1999	All age	Shanghai	Non-invasive	111	Quellung	50.5%	71.2%	70.3%
	Yang F21	1997–1998	≤5 y	Shanghai	Non-invasive	222	Quellung	56.8%	73.9%	60.8%
	Li ZF22	1999–2001	2–14 y	Guangzhou	Non-invasive	84	Quellung	32.1%	46.4%	39.3%
	Zhao GM23	2000-2001	≤3 y	Shanghai	Non-invasive+invasive	112	Quellung	72.3%	81.3%	72.3%
	Wang CQ24	2000–2001	<18 y	Shanghai	Non-invasive	96	Quellung	61.5%	71.9%	62.5%
	Wang CQ25	2000-2002	1m-15 y	Shanghai	Non-invasive	111	Quellung	64.9%	74.8%	64.9%
	Luo XM26	2003	All age	Zhongshan	Non-invasive	266	Quellung	37.6%	53.0%	48.1%
	Sun ZY27	2003-2004	<5	Wuhan	Non-invasive	133	Quellung	5.3%	6.0%	6.0%
	Zhang J28	2004	unknown	Wuhan	Non-invasive	152	Quellung	4.6%	6.6%	7.2%
	Zhang J29	2004	<18 y	Wuhan	Non-invasive	114	Quellung	6.1%	7.0%	7.0%
	Liu YK30	2004-2005	All age	Wuhan	Non-invasive	304	Quellung	4.6%	6.6%	7.2%
	Yang F31	2004–2005	>18 y;<12 y	Shanghai	Non-invasive	103	Quellung, PCR	71.8%	84.5%	86.4%
	Ren HY32	2004–2006	<18 y	Chengdu	unknown	123	Quellung	52.0%	60.2%	68.3%
	Zhang XH33	2004–2009	≥18 y	Hunan	Non-invasive+invasive	822	Quellung	5.1%	20.6%	20.6%
	Pan W34	2005–2012	<18 y	Nanjing	invasive strain	155	Quellung	61.3%	80.6%	85.8%
	Yang T35	2005–2013	All age	Guangdong	Non-invasive+invasive	383	Quellung	16.4%	21.9%	21.9%
	Zhao RZ36	2006–2007	≤5 y	Shenzhen	Non-invasive+invasive	90	Quellung	95.6%	98.9%	100.0%
	Deng QL37	2006-2007	1m一6 y	Guangzhou	Non-invasive	79	Quellung	94.9%	94.9%	97.5%
	Dong YS38	2006–2008	All age	Chongqing	Non-invasive+invasive	143	Quellung	8.4%	10.5%	15.4%
	Dong YS39	2006–2008	unknown	Chongqing	Non-invasive+invasive	133	Quellung	4.5%	4.5%	4.5%
	Zhang H40	2007–2008	≤5 y	Shanghai	Non-invasive	338	Quellung	57.7%	81.1%	80.5%
	Xu F41	2007–2010	<18 y	Nanjing	invasive	48	Quellung	70.8%	93.8%	95.8%
	Miao DQ42	2007–2011	<18 y	Nanjing	Non-invasive+invasive	323	PCR	44.0%	73.7%	62.8%
	Hu JY43	2009	12–18 m	Shanghai	Non-invasive	102	Quellung	63.7%	77.5%	71.6%
	Zhao DF44	2009	12–18 m	Wuhan	Non-invasive	75	Quellung	38.7%	54.7%	46.7%
	Zhang C45	2009–2010	All age	Sichuan	Non-invasive+invasive	166	Quellung	10.2%	12.0%	16.9%
	Li JP46	2009–2010	<18 y	Zhejiang	Non-invasive	106	Quellung	59.4%	95.3%	92.5%
	Ying QH47	2009–2010	All age	Shaoxing	Non-invasive+invasive	103	Quellung	62.1%	62.1%	62.1%
	Zhang B48	2009-2010	All age	Chongqing	Non-invasive+invasive	91	Quellung, PCR	61.5%	75.8%	81.3%
	Chen DL49	2009-2011	All age	Maanshan	Non-invasive+invasive	80	Quellung	55.0%	83.8%	82.5%
	Ma X50	2009-2012	<14 y	Shenzhen	invasive	87	Quellung	72.4%	81.6%	81.6%
	Lin XF51	2009-2013	≥18 y	Wenzhou	invasive	52	Quellung	59.6%	90.4%	86.5%
	Zhou K52	2009-2013	<12 y	Nanjing	invasive	51	Quellung	76.5%	96.1%	98.0%
	Jing CM53	2009-2014	<18 y	Chongqing	Non-invasive+invasive	600	Quellung	53.0%	62.8%	62.8%
	Kang LH54	2010-2013	≤11 y	Chongqing	Non-invasive+invasive	83	Quellung, PCR	51.8%	79.5%	73.5%
	Lu C55	2010-2013	≤14 y	Shenzhen	invasive	76	Quellung	72.4%	94.7%	97.4%
	Liu MJ56	2010-2013	≤11 y	Chongqing	Non-invasive+invasive	46	PCR	52.2%	84.8%	78.3%
	Song XQ57	2010-2014	0-6 y	Taizhou	Non-invasive	322	Quellung	82.9%	95.7%	95.7%
	Li M58	2011-2012	≤5 y	Humen	Non-invasive	229	Quellung	95.2%	97.4%	97.8%
	Huang SY59	2011-2013	All age	Guangzhou	Non-invasive+invasive	94	Quellung	50.0%	67.0%	67.0%
	Ding YF60	2011-2013	< 18 y	Suzhou	Non-invasive	79	PCR	55.7%	93.7%	93.7%
	Jiang QH61	2011-2013	>18 y	Shanghai	Non-invasive+invasive	37	Quellung, PCR	37.8%	48.6%	51.4%
	Wang JR62	2011-2014	0-107m	Wenling	unknown	284	Quellung	60.6%	85.9%	80.6%
	Wang LM63	2011-2015	0-6 y	Jinhua	Non-invasive	302	Quellung	83.8%	97.7%	97.7%
	Li ST64	2012	1-10 y	Nanjing	Non-invasive	584	Quellung	64.7%	92.3%	96.4%
	Fen P65	2012	0-14 y	Shanghai	Non-invasive	328	Quellung, PCR	58.2%	84.1%	77.4%
	Yuan YT66	2012-2013	<7 y;>50 y	Xiamen	Non-invasive+invasive	265	Quellung	62.6%	77.4%	67.5%
	Geng Q67	2012-2013	≤5 y	Suzhou	Non-invasive	175	PCR	66.3%	84.6%	84.6%
	Fen P68	2013	≤14 y	Shanghai	Non-invasive+invasive	284	Quellung, PCR	58.1%	81.0%	79.9%
	Peng YJ69	2015	≤12 y	Chongqing	Non-invasive	267	Quellung	49.1%	72.7%	66.7%
North
	Yu SJ70	1997	≤5 y	Beijing	Non-invasive	190	Quellung	31.6%	44.7%	31.6%
	MCGEE L 71	1997-1998	≤5 y	Beijing	Non-invasive	376	Quellung	47.1%	74.7%	55.1%
	Li J72	1999	≤5 y	Beijing	Non-invasive	97	Quellung	8.2%	10.3%	11.3%
	Yao KH73	2000-2002	<5 y	Beijing	Non-invasive	63	Quellung	4.8%	4.8%	6.3%
	Yao KH74	2000-2004	≤5 y	Beijing	Non-invasive	129	Quellung	7.0%	7.0%	7.8%
	Yu S75	2000-2005	≤5 y	Beijing	Non-invasive	519	Quellung	8.3%	8.5%	11.6%
	Tong YJ76	2004	≤5 y	Beijing	Non-invasive	137	Quellung	10.2%	10.2%	14.6%
	Wu JX77	2008-2010	≤6 y	Linyi	Non-invasive	124	Quellung	29.8%	48.4%	71.8%
	Liu CF78	2009-2011	3m-5 y	Shenyang	invasive	61	Quellung	52.5%	93.4%	100.0%
	Zhou L79	2010	≤5 y	Beijing	Non-invasive	140	Quellung	37.1%	44.3%	37.1%
	Wang TW80	2010-2013	≤5 y	Dalian	Non-invasive	131	PCR	60.3%	60.3%	60.3%
	Cao XH81	2012	<18 y	Luohe	Non-invasive+invasive	134	Quellung	48.5%	48.5%	59.0%
	Li J82	2012-2014	<18 y	Beijing	Non-invasive+invasive	103	Quellung	59.2%	86.4%	80.6%
	Liu ZW83	2013-2014	≤3 y	Taian	Non-invasive+invasive	320	Quellung	75.0%	82.8%	75.0%
	Dong F84	2013-2014	<18 y	Beijing	Non-invasive+invasive	258	Quellung	43.0%	65.5%	64.0%
	Yao KH85	2013-2014	≤14 y	Beijing	Non-invasive+invasive	187	Quellung	25.7%	55.1%	46.0%
	Yu SJ86	2013-2014	≤16	Beijing	Non-invasive	100	Quellung	39.0%	51.0%	44.0%
	Zhou MJ87	2013-2015	All age	Baoding	Non-invasive+invasive	210	PCR	35.2%	62.9%	51.9%
	Wang Q88	2013-2016	3m一7.5 y	Beijing	invasive	30	Quellung	56.7%	96.7%	93.3%
	Wang JF89	2014	<18 y	Zhengzhou	Non-invasive+invasive	134	Quellung	48.5%	48.5%	59.0%
	Wang YT90	2014	≤8 y	Hebei	invasive	43	PCR	48.8%	65.1%	67.4%
	Zhang M91	2014-2015	<18 y	Jinan	Non-invasive	42	Quellung	9.5%	9.5%	23.8%
Multicenter
	Yao KH92	2000-2002	<5 y	Multicenter	Non-invasive	625	Quellung	6.6%	6.6%	6.6%
	Li MC93	2004-2009	unknown	Multicenter	unknown	144	Quellung	8.3%	9.0%	10.4%
	Li MC94	2004-2011	unknown	Multicenter	Non-invasive+invasive	241	Quellung	58.9%	76.8%	74.3%
	Liu YD95	2005-2006	≤4 y	Multicenter	Non-invasive+invasive	451	Quellung	63.6%	75.4%	75.4%
	Xiao SK96	2005-2008	All age	Multicenter	Non-invasive+invasive	580	Quellung	31.7%	52.6%	52.6%
	Liu CL97	2005-2008	All age	Multicenter	invasive	148	Quellung	36.5%	67.6%	66.2%
	Zhao CJ98	2005-2011	All age	Multicenter	Non-invasive+invasive	240	Quellung	34.6%	63.8%	63.8%
	Yao KH99	2006-2007	≤5 y	Multicenter	Non-invasive+invasive	279	Quellung	81.4%	92.8%	93.9%
	Yao KH100	2006-2008	≤5 y	Multicenter	Non-invasive+invasive	338	Quellung	74.0%	87.9%	87.9%
	Ma X101	2006-2008	<14 y	Multicenter	invasive	171	Quellung	63.2%	80.7%	80.7%
	Zhang YJ102	2007-2011	All age	Multicenter	Non-invasive+invasive	39	Quellung	64.1%	87.2%	89.7%
	Chen C103	2008-2010	unknown	Multicenter	Non-invasive+invasive	277	Quellung	45.5%	51.6%	64.6%
	Wang Q104	2010-2011	All age	Multicenter	Non-invasive+invasive	471	Quellung	39.1%	67.5%	62.6%

Notes:

^aVaccine Coverage Rate was calculated as the percentage of isolates from each study that belonged to the serogroups/serotypes included in the PCVs.

Figure 2.

The regional distribution of enrolled studies and pneumococcal isolates.

Of the included articles, 72 articles used the capsular Quellung test for serotyping, 7 used multiplex PCR, 6 used both capsular Quellung test and multiplex PCR; 58 articles only included isolates from children, and 3 articles only reported isolates from adults; 12 articles studied the invasive isolates, and 32 reported both invasive and non-invasive isolates.

Pneumococcal serotypes

Pneumococcal isolates

In total, 16,945 S. pneumoniae isolates were eventually included in this meta-analysis; 11,987 (70.7%) isolates were from children younger than 18 years, 1,963 (11.6%) isolates were from adults, and the other 2,995 isolates were from children or adults. There were 896 (5.3%) invasive isolates, 7,577 (44.7%) non-invasive isolates (the stratification analysis on invasive / non-invasive isolates were only done for children), and the other 8,503 isolates could not be stratified. Stratifying by geographical regions, 9,442 (55.7%) strains were from the south, 3,608 (21.3%) strains were from the north, and the other 3,895 isolates were from multicenter that could not be stratified.

Serotype distribution and vaccine serotype coverage

Of all the included isolates, the most common serotypes were 19F, 19A, 23F, 14, 6B, 6A, 3, 15B, 9V, and 5. The main difference of serotype distribution between children and adults was that serotype 3 was more prevalent in adults than in children. For those isolated from sterile sites in children, the most common serotypes/serogroups were 19F, 19A, 14, 23F, and 6B. Although serotype 19A and serotype 14 were more prevalent in invasive strains than in non-invasive strains, serotype 6A in invasive strains was far less than in non-invasive strains.

Fifty-five point seven percent of the included isolates were from the south of mainland China, and 21.3% were from the north. Serotype 19A and serotype 6B were more prevalent in south region than in north region, though fewer strains of serotype 6A in south region. (Table 2).

Table 2.

Serotype distribution and vaccine-serotype coverage of the pneumococci isolated from 2000 to 2016 in mainland China.

Strains	Serotype /Serogroup	Number (n)	Proportion (%)	95% CI	I2(%) a	P b
Strains from children<18y (N = 11,711)
	19F	3487	29.8	28.3–30.0	/	/
	19A	1192	10.2	9.3–10.4	/	/
	23F	1089	9.3	8.6–9.7	/	/
	14	859	7.3	7.9–8.9	/	/
	6B	581	5.0	6.8–7.7	/	/
	6A	500	4.3	3.9–4.6	/	/
	19	446	3.8	3.5–4.2	/	/
	23	280	2.4	2.1–2.7	/	/
	6	277	2.4	2.1–2.6	/	/
	15	207	1.8	1.5–2.0	/	/
	PCV10	/	52.3	44.3–60.3	99.1	<0.001
	PCV13	/	68.4	60.8–76.	99.5	<0.001
	PPSV23	/	65.5	58.0–73.0	99.4	<0.001
Strains from adults≥18y (N = 1,963)
	19	248	12.6	11.2–14.1	/	/
	3	242	12.3	10.9–13.8	/	/
	6	178	9.1	7.8–10.3	/	/
	19F	172	8.8	7.5–10.0	/	/
	23	146	7.4	6.3–8.6	/	/
	19A	122	6.2	5.1–7.3	/	/
	14	106	5.4	4.4–6.4	/	/
	23F	71	3.6	2.8–4.4	/	/
	15	52	2.6	1.9–3.4	/	/
	17	25	1.3	0.8–1.8	/	/
	PCV10	/	29.7	19.2–40.3	97.0	<0.001
	PCV13	/	49.5	32.6–66.3	98.6	<0.001
	PPSV23	/	50.7	34.6–66.9	98.3	<0.001
Invasive strains (N = 896)
	19F	210	23.5	20.6–26.3	/	/
	19A	201	22.2	19.4–25.0	/	/
	14	146	16.4	13.9–18.9	/	/
	23F	59	6.7	5.0–8.4	/	/
	6B	44	5.1	3.6–6.6	/	/
	9V	32	3.6	2.4–4.8	/	/
	1	17	1.9	1.0–2.8	/	/
	8	15	1.7	0.8–2.5	/	/
	7F	13	1.5	0.7–2.2	/	/
	5	10	1.1	0.4–1.8	/	/
	PCV10	/	61.4	55.2–67.7	70.6	<0.001
	PCV13	/	87.4	82.7–92.0	79.4	<0.001
	PPSV23	/	86.9	81.7–92.2	84.6	<0.001
Non-invasive strains (N = 7,577)
	19F	2,006	26.5	25.5–27.5	/	/
	19A	637	8.4	7.8–9.0	/	/
	23F	600	7.9	7.3–8.5	/	/
	14	499	6.6	6.0–7.1	/	/
	19	439	5.8	5.3–6.3	/	/
	6A	371	4.9	4.4–5.4	/	/
	6B	350	4.6	4.1–5.1	/	/
	23	299	3.9	3.5–4.4	/	/
	6	241	3.2	2.8–3.6	/	/
	15	171	2.3	1.9–2.6	/	/
	PCV10	/	43.2	33.2–53.2	99.4	<0.001
	PCV13	/	55.7	43.5–67.8	99.7	<0.001
	PPSV23	/	53.4	41.5–65.3	99.7	<0.001
Strains from the south of China(N = 9,442)
	19F	2452	26.0	25.1 -26.9	/	/
	19A	852	9.0	8.4 -9.6	/	/
	23F	811	8.6	8.0 -9.2	/	/
	19	747	7.9	7.4–8.5	/	/
	14	609	6.4	6.0–6.9	/	/
	6B	452	4.8	4.4–5.2	/	/
	6	382	4.0	3.6–4.4	/	/
	6A	348	3.7	3.3–4.1	/	/
	23	312	3.3	2.9–3.7	/	/
	3	237	2.5	2.2–2.8	/	/
	PCV10	/	50.1	40.5–59.6	98.2	<0.001
	PCV13	/	64.5	54.8–74.2	99.3	<0.001
	PPSV23	/	62.8	53.2–72.4	98.7	<0.001
Strains from the north of China (N = 3,608)
	19F	508	14.1	12.9–15.2	/	/
	23F	343	9.5	8.5–10.5	/	/
	19	289	8.0	7.1–8.9	/	/
	14	264	7.3	6.5–8.2	/	/
	6A	233	6.5	5.7–7.3	/	/
	19A	191	5.3	4.6–6.0	/	/
	23	143	4.0	3.3–4.6	/	/
	6	113	3.1	2.6–3.7	/	/
	6B	105	2.9	2.4–3.5	/	/
	15	92	2.5	2.0–3.1	/	/
	PCV10	/	36.3	26.7–45.9	98.2	<0.001
	PCV13	/	50.5	36.2–64.8	99.3	<0.001
	PPSV23	/	47.8	35.7–59.9	98.7	<0.001

Notes:

^aI² means the total variation across studies that were due to heterogeneity.

^bP: The Cochran chi-square (χ²) test of Heterogeneity. A P value of <0.05 was considered statistically significant.

/: Not applicable.

During 1996–1999, the most prevalent serotypes/serogroups were 23F, 6A, 19F, 14, and 6B; during 2005–2009 and 2010–2016, the most prevalent serotypes/serogroups were 19F, 19A, 23F, 14, and 6B. The major changes were the increased frequency of serotype 19A and 19F since 2000–2004. Fig. 3 shows the distribution of vaccine serotypes of pneumococcal isolates from 1996 to 2016 in mainland China.

Figure 3.

Distribution of vaccine serotypes of pneumococcal isolates from 1996 to 2016 in mainland China.

The range of pooled coverage for PCV10 serotypes was from 29.7% to 62.3%, that for PCV13 was 43.2%–61.4%, and that for PPSV23 was 50.7%–65.5%, depending on different age groups. For different sources, the pooled coverage for PCV10 serotypes ranged from 43.2% to 61.4%, that for PCV13 ranged from 55.7% to 87.4%, and that for PPSV23 ranged from 62.8% to 86.9%. Regarding to different regions, the pooled coverage for PCV10 serotypes varied from 36.3% to 50.1%, that for PCV13 varied from 50.5% to 64.5%, and that for PPSV23 varied from 47.8% to 62.8%. (Table 2) Before 2004, the pooled coverage rates for vaccines serotypes were very low, which were 24.9% (95% CI: 14.8%–35.0%) for PCV10 serotypes, 29.6% (95% CI: 16.2%–43.0%) for PCV13 and 27.1% (95% CI: 15.8%–38.4%) for PPSV23. Since 2005, the pooled coverage for vaccines serotypes increased gradually. During 2010–2016, the pooled coverage for PCV10 serotypes was 54.4% (95% CI: 46.5%–62.3%), that for PCV13 was 73.3% (95% CI: 67.8%–78.8%) and that for PPSV23 was 71.6% (95% CI: 65.9%–77.2%). (Fig. 3)

Heterogeneity and meta-regression

We noticed high heterogeneity between studies (I² = 70.6%–99.7%) (Table 2). We performed meta- regression analysis and used variables, including age (<18 y vs ≥18 y) region (northern vs southern), source (invasive vs non-invasive), serotype method (PCR vs Quellung), number of strains, and study year. In univariate analysis, we noticed that regions (p = 0.003) and study year (p = 0.016) were associated with the coverage rate of PCV13 serotypes. However, in multivariable analysis, regions, sources and study years explained 41% of the total heterogeneity. (Table 3)

Table 3.

Results of Meta-regression for pneumococcal serotype coverage of PCV13 in mainland China from 2000 to 2016.

Covariatea	Meta-regression Coefficient	95% Confidence Interval	P value	Adj R-squaredb
Univariate analyses
Age (<18y vs ≥18y)	–0.268	–0.944 to 0.407	0.427	–0.92%
Region (northern vs southern)	0.355	0.133 to 0.578	0.003	25.37%
Source (invasive vs non-invasive)	0.273	–0.142 to 0.689	0.188	2.82%
Serotype (PCR vs Quellung)	–0.077	–0.399 to 0.246	0.629	–2.70%
Number of strains	0.000	–0.001 to 0.001	0.663	–2.13%
Study year	–0.515	–0.865to –0.166	0.005	35.80%
Multivariable analyses				41.18%
Age (<18y vs ≥18y)	0.076	–0.474 to 0.626	0.780
Region (northern vs southern)	0.345	0.174 to 0.516	<0.001
Source (invasive vs non-invasive)	0.331	0.021 to 0.641	0.037
Serotype (PCR vs Quellung)	–0.142	–0.385 to 0.100	0.241
Number of strains	0.001	–0.000 to 0.001	0.092
Study year	–0.370	–0.665 to -0.076	0.016

Notes:

^aThe dependent variance of the Meta-regression was the serotype coverage rate (r) of PCV13. The covariates included in the univariate and multivariable models were age (<18 y vs ≥18y), region (northern vs southern), source (invasive vs non-invasive), serotype (PCR vs Quellung), number of strains, and study year.

^bAdj R-squared was the proportion of variance that can be explained.

Discussion

Of the currently available data from the published papers, serotype 19F pneumococcus was the most common cause of invasive and non-invasive pneumococcal disease (PD) in mainland China, followed by serotype 19A pneumococcus. Among children aged <18 years, serotype 19F pneumococcus was the main cause of IPD, a finding that was different from Johnson's finding that serotype 14 pneumococcus was the most common cause of IPD in Africa, Asia, Europe, and North America etc.⁸ Thus, the most common serotypes causing IPD may vary slightly across geographic regions.

Among the strains isolated from children aged younger than 18 years, serotypes 19F, 19A and 23F were the most common serotypes, results that were similar to those of Lyu et al.⁷ In addition, serotype 23F, 14 and 6B were important causes of either PD or IPD in children. This finding was consistent with the results from Western Europe and the USA that serogroups 14, 23, and 6 were common serogroups.⁹ However, among the strains isolated from adults, the most common serotypes/serogroups were 19, followed by 3, and 6. Similarly, several studies in Asia reported that the prevalence of serotype 3 increased from 14.3% to 24.3 % in adults and even became the most prevalent among patients aged older than 65 years.^10,11 Thus, the prevalent serotypes may vary among age groups. The most prevalent serotypes were similar in the north and south, and serotype/serogroup 19F (or 19) was the most frequent in the various regions, though the rate of each serotype is different. Additionally, serotype19A is more frequent in the south than in the north. Thus, the serotypes of S. pneumonia vary in different areas.

For all of the isolates included in our study, the coverage rates of PCV13 and PPSV23 were significantly higher than those of PCV10, with no significant differences between PCV13 and PPSV23, a finding that was s consistent with other study findings in China.^7,12 Although the number of serotypes in PPSV23 was more than that in PCV13, the serotype coverage of PPSV23 was slightly lower than that of PCV13, potentially due to the lack of serotype 6A in PPSV23 and the contribution of serotype 6A is higher than the combined contribution of the 11 serotypes unique to PPSV23. PCV13 has higher coverage compared with PCV7,¹³ and the additional serotypes 1, 5, 7F, 3, 6A and 19A, especially for serotype 19A, account for a large proportion. In our study, PCVs or PPSV coverage among children was higher than that among adults, consistent with the results of a study from China.¹⁴ Among the isolates from children, PCVs or PPSV coverage of invasive isolates was higher than that of non-invasive isolates. In the USA, the rate of IPD reduced dramatically after 5 years since the license of PCV13, especially driven by decreases in 19A.¹⁵ Considering the comprehensive evidence concerning the effectiveness of PCV13 we could expect a similar reduction of IPD if we have a high vaccination rate of PCV13 in mainland China.

In mainland China, PCV7 was licensed since 2008 and PCV13 was licensed since 2016, but both as self-paid vaccine and the extent of adoption was very low. Thus the natural fluctuation in serotype distribution in China for the last 20 years appears to not have significantly influenced from PCVs introductions. The major changes were the increasing of serotype 19A and 19F since 2000–2004, which may have been mainly due to the selective pressure of antibiotics usage. In the USA, the non-PCV7 serotype 19A increased in prevalence from 2.7% in 1999–2000 to 34.1% in 2010–2011 after 10 years of PCV7 licensed, but these serotype fluctuations have been mainly attributed to serotype replacement following PCV7 immunization.^16,17

In other countries, serotype shift is a phenomenon since the use of PCV7.^13,14 Although PCV7 is not widely used in China, the increasing of serotype19A was obvious since 2000, and serotype 19A is one of the most common serotypes in most areas of China.¹⁵ The serotype shift in China may be due to the selective pressure of antibiotics rather than the use of PCV7, because of the antibiotics were commonly used in China. In the future, we need to keep monitoring the serotype distribution and change.

There are some limitations in this systematic review and meta-analysis. First, we used different thresholds for a number of invasive (>30) and of non-invasive isolates (>40), which might lead to some bias. In consideration of the common usage of antibiotics, lower bacterial load and the quality of the specimens, it's more difficult to isolate the invasive pneumococcal strains in the fields, but the invasive pneumococcal diseases are more severe than non-invasive pneumococcal diseases. Therefore, we used a lower threshold for invasive isolates to include more meaningful data in this meta-analysis. Second, we observed large heterogeneity of the included articles, but regions, sources, and study years only explained part of the observed heterogeneity, other possible reasons remain unknown. Third, serotype coverage of vaccines may be underestimated in our study, because only serotypes were used in the pooled coverage regardless of serogroup data. Furthermore, most of the included studies came from southeastern of mainland China; few came from northwestern provinces, such as Xinjiang, Qinghai and Ningxia province. To better understand the impact of PCV13 on these areas, more data are needed.

In conclusion, pneumococcal serotype 19F was the most prevalent serotype in mainland China, and the most common serotypes varied among the population, between invasive and non-invasive isolates, and across geographic regions. The serotype coverage rates of PCV13 and PPSV23 were high in mainland China. During and after the introduction of PCV13 in China, more scientific data are required, especially from northwestern China; the serotype distribution should be monitored over an extended time period.

Methods

Literature search

A systematic literature search was conducted in PubMed, Ovid-EMBASE, Web of Science, CNKI, and Wanfang. We searched electronic medical databases for studies (published between January 1, 2000 and December 31, 2016) reporting pneumococcal serotypes from hospitalization patients or isolates from healthy people in mainland China. Strategies were designed to retrieve records that included the following terms: (“Streptococcus pneumoniae” OR “S. pneumoniae” OR “pneumococcus” OR “pneumococci” OR “diplococcus pneumoniae” OR “diplococcus” OR “pneumococcal”) and (“serotype” OR “serotypes” OR “serogroups” OR “serogroup”) and (“China” OR “Chinese”)

Criteria for selection

We included an article if (1) It was an original study, and the isolates were serotyped by multiplex polymerase chain reaction or capsular Quelling reaction methods; (2) the number of isolates was ≥30 for invasive isolates or ≥ 40 for non-invasive isolates; and (3) the strains were isolated from the population in mainland China (except Hong Kong, Macao and Taiwan). We excluded an article if (1) it was a review, case report or lecture; (2) it lack a detailed description about the serotypes of the isolates; (3) it was a duplicate publication. For more detailed information, please refer to Fig. 1.

Data extraction

The following information was extracted from each selected paper: first author's name, study region, age of study population, source of isolates, years of isolates collected, number of isolates, and the serotypes.

Invasive pneumococcal isolates were defined as S. pneumoniae strains identified from a normally sterile site (e.g., blood, cerebrospinal, pleural effusions, or joint fluid, etc.). Non-invasive pneumococcal strains were defined as S. pneumoniae strains isolated from non-sterile sites such as sputum or middle ear effusion. Vaccine serotypes referring to the serotypes included in the pneumococcal conjugate vaccines or pneumococcal polyvalent vaccine (details in Box 1).

Box 1. Serotypes in Pneumococcal conjugate vaccines or polysaccharide vaccine

PCV7

4

6B

9V

14

18C

19F

23F

PCV10

4

6B

9V

14

18C

19F

23F

1

5

7F

PCV13

4

6B

9V

14

18C

19F

23F

1

5

7F

3

19A

6A

PPSV23

4

6B

9V

14

18C

19F

23F

1

5

7F

3

19A

2

8

9N

10A

11A

12F

15B

17F

20

22F

33F

Data analysis

The collected isolates were stratified by the properties of each study into children or adults, invasive isolates or non-invasive isolates, southern or northern regions. Southern or northern regions were defined by Qinling Mountains and Huai River.

I² was used to describe the percentage of the total variation across studies that were due to heterogeneity. I² = 25% was associated with low heterogeneity, I² = 50% was associated with moderate heterogeneity, and I² = 75% was associated with high heterogeneity.¹⁸ Two-sided p ≤ 0.05 was considered statistically significant.¹⁹ Most of our meta-analysis, showed high heterogeneity (I²>75%). Considering the high heterogeneity among studies, we used the random-effects models for the pooled estimation of the serotype coverage rate of PCVs.

To explore which study characteristics explained the large heterogeneity, meta-regression was conducted according to potentially relevant characteristics.

The PCV serotype coverage rate (r) was calculated as the total number of isolates included in the vaccine serotypes divided by the total number of isolates that were serotyped. The standard error (SE) of the coverage rate was calculated as SE = Sqr(r*(1-r)/n). The coverage rate of PCVs from different studies were pooled with using the random-effects model.

All statistical analyses were performed using Stata SE 12.0 (Stata Corp, College Station, TX).

Funding Statement

This study was supported by the National Natural Science Foundation of China [81102166], the Shanghai Municipal Health and Family Planning [GWTD2015S05 and 15GWZK0101], the National Key Research and Development Program of China (2017YFC0211700) and the SINO-US collaborative program on Emerging and Re-Emerging Infectious Diseases [5U2GGH000018].

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.