Table 1 Plasmid characteristics for plasmids encoding/not encoding any of the 10 major ARG types.

From: Factors associated with plasmid antibiotic resistance gene carriage revealed using large-scale multivariable analysis

Explanatory variables

Measures

 ≥ 1 major ARG type

(n = 3639)

No major ARG type

(n = 10,504)

log10 Plasmid size*

4.85 [4.59, 5.11]

4.60 [3.95, 4.99]

Insertion sequence density (frequency per 10 kb)*

0.40 [0.23, 0.63]

0.04 [0.00, 0.35]

Collection date (years since initial collection year [1994])*

21.00 [18.00, 23.00]

18.00 [14.00, 21.00]

Integron present

1232 (33.9)

9 (0.1)

Biocide/metal resistance gene present

1486 (40.8)

572 (5.4)

Virulence gene present

226 (6.2)

933 (8.9)

Conjugative system

 Non-mobilisable

847 (23.3)

5533 (52.7)

 Mobilisable

823 (22.6)

3110 (29.6)

 Conjugative

1969 (54.1)

1861 (17.7)

Replicon carriage†

  

 Untyped

637 (17.5)

7593 (72.3)

 Single-replicon

1995 (54.8)

2226 (21.2)

 Multi-replicon

1007 (27.7)

685 (6.5)

Host taxonomy

 Enterobacteriaceae

2558 (70.3)

2072 (19.7)

 Proteobacteria (non-Enterobacteriaceae)

593 (16.3)

3767 (35.9)

 Firmicutes

391 (10.7)

2543 (24.2)

 Other

97 (2.7)

2122 (20.2)

Geographic location††

 High-income not elsewhere classified

257 (7.1)

1430 (13.6)

 Middle-income not elsewhere classified

183 (5.0)

772 (7.3)

 European Union (EU) and United Kingdom

172 (4.7)

1070 (10.2)

 China

286 (7.9)

962 (9.2)

 United States

351 (9.6)

1140 (10.9)

 Other

2390 (65.7)

5130 (48.8)

Isolation source

 Human

1024 (28.1)

1930 (18.4)

 Livestock

152 (4.2)

556 (5.3)

 Other

2463 (67.7)

8018 (76.3)

  1. Note: Showing median [IQR] for continuous variables and n (%) for categorical variables. The number of other antibiotic resistance gene (ARG) types is not shown as a continuous variable since it was determined per ARG type model rather than for the overall dataset. The number of other ARG types was winsorised at the modal 95th percentile which corresponded to four other ARG types (the 95th percentile for the colistin model was five other ARG types, and the 95th percentile for all other models was four other ARG types). “Other” factor levels include plasmids assigned to a rare category or not assigned a category (if BioSample metadata was missing; or for isolation source, non-human/livestock metadata was left un-curated and assignment to specific categories was not attempted). For host taxonomy, taxonomic metadata was present for all plasmids, but sometimes only “uncultured bacterium” (n = 261); for geographic location, metadata was missing for 7284 plasmids; for isolation source, metadata was missing for 4317 plasmids.
  2. *Continuous variables were winsorised (limiting extreme values to threshold values) at the 5th/95th percentiles on both tails (plasmid size, 2.9–311.7 kb); right tail (insertion sequence density, 1.05 per 10 kb); left tail (collection date, 1994).
  3. †For replicon carriage, the factor level reflects the number of unique replicon types detected (e.g. IncFIB, IncFIC type is categorised multi-replicon whereas IncFIC, IncFIC is categorised single-replicon).
  4. ††For geographic location, the middle-income not elsewhere classified (n.e.c.) category represents World Bank lower-middle income and upper-middle income countries combined, and not elsewhere classified. China is an upper-middle income category; the United Kingdom and United States are high-income countries, and European Union countries are all high-income except for Romania (upper-middle income) (income categorisations based on World Bank 2018 income groupings).
Back to article page