After reading abstracts and reviewing the full text, 33 studies (26 – India, 5 – Bangladesh, 2 – Pakistan) fulfilled the a priori selection criteria and were included in the meta-analysis ( Table 1). Fourteen of the titles represented recent data not available in past reviews [18], [37] and [63] and included studies using more advanced molecular methods for strain characterization. Both frontline urban hospitals and rural community health centers served as surveillance sites for collecting samples. Studies characterized both symptomatic

and asymptomatic rotavirus cases from rainy and dry seasons. A large variation in laboratory methods to detect rotavirus types was observed, with earlier studies (before 1994) relying principally on ELISA and PAGE, and later studies utilizing more advanced molecular RT-PCR techniques. Prior to 1994, two studies check details utilized PAGE, two utilized ELISA, and three utilized RT-PCR. From 1995 to 1999, 11 studies were published with 4 reporting PAGE techniques and 6 reporting RT-PCR; one study did not specify laboratory methods. The 15 studies from 2000 to 2009 relied entirely upon RT-PCR

for genotyping, which represents the first time period that all results were fully based on RT-PCR techniques. Overall, due to their later discovery in humans, 25 of the 33 studies (76%) did not use typing agents for detection of G12 while 11 of the earlier studies (33%) did not determine the G9 type. This is reflected in the proportion of “untypeable” strains that were NVP-AUY922 purchase observed. When untyped strains were considered in the denominator of all tested specimens, 23.7% were untypeable prior to 2000. However, after 2000, when molecular typing methods were used and included primers for the G9 and G12 strains, the proportion of untypeable strains was reduced to 13.7%. A similar trend was noted in the results for the VP4 P-type, where 21.3% of strains could not be typed before 2000, compared to 16.3% after 2000, probably due to the wider range of primer sets used. The 33 studies provide data on 9,153 rotavirus samples examined for the VP7 G-type, while 21 studies present results

for 4,842 VP4 P-types. Among typeable G-samples (n = 7703) over the period covered in this review (1983–2009), the four most globally Thymidine kinase common types, G1 (31.4%), G2 (29.4%), G3 (3.6%), and G4 (13.8%), represented approximately 78% of total samples. During this same time period, G9 (11.2%), G-Mixed (6.9%), and G12 (3.7%) were also identified ( Table 2). For the P-types, between 1983 and 2009, P[4] (29.3%) and P[8] (44.7%) represented approximately 75% of all the 4148 typeable P-strains, with P[6] (15.2%) and P-Mixed (10.8%) also present ( Table 3). However, the percentages of uncommon G-types and mixed P-types reported may not accurately reflect the true proportions circulating in the population due to the number of untypeable strains showing current techniques.