Background:
We compared the hospital-based epidemiology of neonatal sepsis after the coronavirus disease 2019 lockdown (LD) versus historical epochs and the LD period versus phases of unlocking.
Methods:
This retrospective cohort study was conducted in a level 3 neonatal unit. We compared neonates born in three 24-week periods—Group LD: 22 March 2020 to 5 September 2020—the reference group, Group pre-LD: 29 September 2019 to 14 March 2020 and Group temporally corresponding to LD in 2019 (corres-LD): 24 March 2019 to 7 September 2019. We also studied linear trends from LD phase 1.0 until Unlock 4.0. The key outcome was culture-positive sepsis.
Results:
There were 1622, 2744 and 2700 subjects in groups LD, pre-LD and corres-LD, respectively. The incidence of any culture-positive sepsis in pre-LD was higher than LD [odds ratio (95% CI) = 1.61 (1.02–2.56)]. This was mainly due to a statistically significant reduction in Acinetobacter baumannii sepsis, with incidence rate differences of pre-LD versus LD [0.67 (95% CI: 0.37–0.97), P = 0.0001] and corres-LD versus LD [0.40 (95% CI: 0.16–0.64), P = 0.0024]. Groups pre-LD and corres-LD had higher proportion of multi-drug resistant (MDR)/extreme drug resistance/pan drug resistance sepsis than LD [77%, 77% and 44%, respectively (P values of both groups vs. LD = 0.01)]. From LD 1.0 to unlock 4.0, there were fewer episodes of MDR sepsis (Plinear trends = 0.047). On multivariable analysis, group pre-LD (vs. reference group LD), male sex, birth weight and Apgar score independently predicted culture-positive sepsis.
Conclusions:
LD favorably impacted the epidemiology of neonatal sepsis in a hospital setting, with less A. baumannii and MDR sepsis, which persisted during unlocking.
Keywords: SARS-CoV-2, Acinetobacter baumannii, septicemia, multi-drug resistant, newborn
The coronavirus disease 2019 (COVID-19) pandemic has directly afflicted millions globally.1 The pandemic has also caused collateral damage from other diseases.2–4 A recent meta-analysis identified increased stillbirths, maternal deaths and postnatal depression during the pandemic.5 Conversely, COVID-19–related restrictions have decreased the incidence of several non–COVID-19 viral diseases.6–11 Respiratory viral infections decreased after the pandemic, despite more testing for viral illnesses.12 To date, there are no reports regarding the effect of the COVID-19 pandemic on the changing epidemiology of neonatal sepsis.
Ours is a level III Neonatology Unit in a public-sector university hospital in northwest India. We deliver >6000 liveborn infants annually, of which ≈40% are preterm and ≈35% small for gestational age. Like other busy public-sector units in developing countries, neonatal sepsis is a major problem. 2%–3% of neonates in our unit develop culture-positive sepsis (unpublished 5-year unit data). Etiologic organisms include Acinetobacter baumannii, Escherichia coli, Klebsiella pneumoniae and coagulase-negative Staphylococcus, in that order. Multi-drug resistant (MDR) sepsis is common, like other parts of South Asia.13
In the absence of regionalized care, our center is overburdened with sick neonates. We have a 16-bedded neonatal intensive care unit (NICU) and a 24-bedded high dependency unit (HDU). We are often unable to immediately transfer neonates requiring NICU or HDU care to the respective facilities, owing to nonavailability of vacant beds, nor can we transfer out such patients. Consequently, we manage several neonates in the labor room stabilization unit- an interim high-dependency nursery- until such time that a transfer to the NICU or HDU is feasible. A surge in the delivery rate is a surrogate for overcrowding, and we have earlier anecdotally observed an increase in sepsis rates with overcrowding.
The government of India announced a public curfew on March 22, 2020, severe restrictions on March 23–24 and a stringent lockdown (LD) starting from March 25, 2020. The LD was extended until May 31, followed by unlocking in 1-month phases, during which restrictive measures were eased.
As expected, there were fewer neonates delivered in our center from March 22, 2020 onwards. A few months into the pandemic, we observed a decrease in episodes of culture-positive sepsis. In the absence of formal analysis of data, we were unsure about the validity of this observation and whether it was a consequence of fewer admissions, behavioral changes, administrative measures or a combination thereof. Improvements in the pattern of neonatal sepsis may hold useful lessons for a future non-pandemic situation.
Hence, we studied neonates delivered in 3 time periods, each lasting 6 months—the first starting with the LD in 2020, the second immediately prior to the LD and the third in 2019, temporally identical to the LD. We investigated trends across the phases of locking and unlocking. We hypothesized that the LD was associated with a lower incidence of neonatal sepsis, especially MDR sepsis and that unlocking was associated with a partial reversal of the same.
METHODS
We conducted a retrospective cohort study on neonates delivered in our hospital in three 24-week periods (see Figure, Supplemental Digital Content 1, http://links.lww.com/INF/E680):
Group LD: 22 March 2020 to 5 September 2020.
Group pre-LD: prior to the LD: 29 September 2019 to 14 March 2020.
Group temporally corresponding to LD in 2019 (corres-LD): 24 March 2019 to 7 September 2019.
Group LD was the reference group. We included Group corres-LD to control for the possible impact of seasonal variations on demographics and sepsis.
To study the effect of stringency of measures, we included all liveborn infants delivered from 22 March 2020 to 30 September 2020 (see Figure, Supplemental Digital Content 1; http://links.lww.com/INF/E680). A composite objective measure of restrictions is the Government Stringency Index, ranging from 0 to 100, with 100 representing maximum stringency.14 This part of the study included the following phases:
| Phase of LD/Unlocking | Duration (mm/dd/yyyy) | Government Stringency Index |
|---|---|---|
| LD phases 1 and 2 | March 22, 2020–May 3, 2020 | 96.3–100 |
| LD phases 3 and 4 | May 4, 2020–May 31, 2020 | 79.17–81.94 |
| Unlock 1.0 | June 1, 2020–June 30, 2020 | 75.46–76.39 |
| Unlock 2.0 | July 1, 2020–July 31, 2020 | 74.07–77.78 |
| Unlock 3.0 | August 1, 2020–August 31, 2020 | 79.63 |
| Unlock 4.0 | September 1, 2020–September 30, 2020 | 73.61–81.02 |
The time period of LD/unlocking was almost identical to Group LD, except that the former extended to 30 September 2020, whereas the latter ended on 5 September 2020. Detailed descriptions of the containment measures can be accessed elsewhere.15
We recorded demographic data, admission to the NICU and/or HDU, duration of hospital stay, the final outcome and sepsis data during hospital stay. Antibiotic resistance was defined as “no MDR,” MDR, extreme drug resistance (XDR) and pan drug resistance (PDR).16 In Group LD, we performed a COVID-19 GeneXpert test in all laboring mothers and isolated the neonates from their mothers and other neonates until the mother was reported negative. We recorded the number of neonates thus isolated, and the number of neonates and mothers whose test was positive.
We used χ2 test, Fisher exact test, Mann-Whitney U or Student t test as appropriate. We calculated incidence rate ratios of episodes of culture-positive, A. baumannii and antibiotic-resistant sepsis. We compared the phases of LD/unlocking by tests for linear trends [χ2 for linear trends, Jonkheere-Terpstra or Analysis of Variance for trends, as appropriate] and without linear trends (χ2, Kruskal-Wallis test or Analysis of Variance, as appropriate). We performed a multivariable logistic regression to predict “any episode of culture-positive sepsis,” with the following predictors selected a priori: Groups (Group LD: reference group), number of live births in the week that the subject was born (as a surrogate for overcrowding), sex, gestational age in weeks, birth weight in grams, 5-minute Apgar score and appropriateness for gestation. We tested for multi-collinearity using variance inflation factor and Eigenvalues. We included a constant and forced all variables into the model. We used SPSS v23 and Stata v14.2 for statistical analysis. As recruitment was based on time periods, we did not perform a formal sample size calculation.
RESULTS
There were 1622, 2744 and 2700 inborn neonates in groups LD, pre-LD and corres-LD, respectively (Table 1). Fewer neonates in group pre-LD and corres-LD were admitted in the NICU and HDU and at significantly higher postnatal ages. Neonates with gestational age <32 weeks in group pre-LD and corres-LD had shorter hospital stay. Compared with group LD, a significantly lower proportion in group pre-LD was discharged home alive.
TABLE 1.
Comparison of Demographic, Admission, and Discharge Data Between 3 Groups
| Variable | LD Period: Group “LD”(N = 1622) | Immediate Previous Period: Group “Pre-LD”(N = 2744) | Corresponding Period In 2019: Group “Corres-LD”(N = 2700) | P Value: Group “LD” vs. “Pre-LD” (Reference Group = “LD”) | P Value: Group “LD” vs. “corres-LD” (Reference Group = “LD”) |
|---|---|---|---|---|---|
| Gestational age, weeks | |||||
| Median (1st, 3rd quartile) | 37.0 (34.0, 38.0) | 37.0 (35.0, 38.0) | 37.0 (34.0, 38.0) | 0.1 | 0.022 |
| Mean (SD) | 35.77 (3.2) | 35.90 (3.1) | 35.92 (3.1) | ||
| Gestation categories* | 0.2 | 0.049 | |||
| Extremely preterm (<28 weeks) | 44 (2.7) | 68 (2.5) | 65 (2.4) | ||
| Very preterm (28–<32 weeks) | 125 (7.7) | 220 (8.0) | 230 (8.5) | ||
| Moderate to late preterm (32–<37 weeks) | 574 (35.4) | 880 (32.1) | 844 (31.3) | ||
| Term (37–41 weeks) | 877 (54.1) | 1572 (57.3) | 1554 (57.6) | ||
| Post-term (>41 weeks) | 2 (0.1) | 4 (0.1) | 7 (0.3) | ||
| Birth weight, grams, median (1st, 3rd quartile) | 2480.0 (1896.75, 2874.25) | 2470.0 (1915.0, 2862.0) | 2480.0 (1900.0, 2890.0) | 0.9 | 0.9 |
| Birth weight categories* | |||||
| <1000 g | 59 (3.6) | 129 (4.7) | 123 (4.6) | 0.3 | 0.5 |
| 1000–1499 g | 154 (9.5) | 236 (8.6) | 277 (10.3) | ||
| 1500–2499 g | 599 (36.9) | 1051 (38.3) | 979 (36.3) | ||
| 2500–4200 g | 809 (49.9) | 1324 (48.3) | 1318 (48.8) | ||
| >4200 g | 1 (0.1) | 4 (0.1) | 3 (0.1) | ||
| Sex* | |||||
| Male | 883 (54.4) | 1418 (54.0) | 1426 (52.8) | 0.9 | 0.5 |
| Female | 737 (45.4) | 1259 (45.9) | 1268 (47.0) | ||
| Ambiguous | 2 (0.1) | 4 (0.1) | 6 (0.2) | ||
| Appropriateness for gestational age* | |||||
| AGA | 999 (64.5) | 1713 (63.3) | 1670 (61.9) | 0.3 | 0.2 |
| SGA | 513 (33.1) | 944 (34.9) | 971 (36.0) | ||
| LGA | 37 (2.4) | 50 (1.8) | 57 (2.1) | ||
| Apgar score at 5 min | |||||
| Median (1st, 3rd quartile) | 7 (5, 8) | 7 (6, 8) | 7 (6, 8) | <0.001 | <0.001 |
| Mean (SD) | 7.42 (1.6) | 7.29 (1.6) | 7.23 (1.5) | ||
| Admitted in NICU* | 177 (10.9) | 184 (6.7) | 213 (7.9) | <0.001 [OR, 0.59 (95% CI: 0.47–0.73,)] | 0.001 [OR, 0.70 (95% CI: 0.57–0.86)] |
| Admitted in HDU* | 297 (18.3) | 313 (11.4) | 265 (9.8) | <0.001 [OR, 0.57 (95% CI: 0.48–0.68)] | <0.001 [OR, 0.49 (95% CI: 0.41–0.58)] |
| Age transferred to NICU, d, median (1st, 3rd quartile) | 2 (1, 3) | 3 (2, 5) | 3 (2, 5) | <0.001 | <0.001 |
| Age transferred to HDU, d, median (1st, 3rd quartile) | 7 (4, 13) | 8 (5, 14) | 10 (5, 17) | 0.003 | <0.001 |
| Duration of hospital stay among <32 weeks gestation, d, median (1st, 3rd quartile) | 20 (4, 39) | 10 (2, 30) | 12 (3, 29) | 0.001 | 0.005 |
| Final outcome | N = 1622 | N = 2744 | N = 2694 | ||
| Discharged home | 1553 (95.7) | 2583 (94.1) | 2541 (94.3) | 0.028 | 0.13 |
| Died | 67 (4.1) | 145 (5.3) | 142 (5.3) | ||
| LAMA or DOR | 1 (0.1) | 2 (0.1) | 4 (0.1) | ||
| Transferred elsewhere | 1 (0.1) | 14 (0.5) | 7 (0.3) | ||
Figures in brackets are percentages of the total subjects in the respective period.
AGA indicates appropriate for gestational age; DOR, discharged on request; LAMA, left against medical advice; LGA, large for gestational age; OR, odds ratio; SGA, small for gestational age.
Group pre-LD had a significantly higher proportion of neonates with any episode of culture-positive sepsis compared with group LD [odds ratio = 1.61 (95% CI: 1.02–2.56)], but not so in the case of corres-LD (Table 2). Group pre-LD had a significantly higher incidence rate ratio (IRR) of culture-positive sepsis compared with group LD [1.77 (95% CI: 1.13–2.85)], but not when corres-LD was compared with group LD. There were significant differences in etiologic organisms between the groups. Of note, there were no episodes of A. baumannii sepsis in Group LD, whereas there were 19 (24%) and 11 (20.7%) episodes in groups pre-LD and corres-LD, respectively. The IRR of A. baumannii sepsis in group pre-LD or corres-LD versus LD could not be calculated as the denominator was zero. Hence the incidence rate differences were calculated which were statistically significantly different for pre-LD versus LD [0.67 (95% CI: 0.37–0.97)] and corres-LD versus LD [0.40 (95% CI: 0.16–0.64)].
TABLE 2.
Comparison of Sepsis-Related Data Between 3 Groups
| Variable | LD Period: Group “LD”(N = 1622) | Immediate Previous Period: Group “Pre-LD”(N = 2744) | Corresponding Period in 2019: Group “Corres-LD”(N = 2700) | P Value: Group “LD” vs. “Pre-LD” (Reference Group = “LD”) | P Value: Group “LD” vs. “Corres-LD” (Reference Group = “LD”) |
|---|---|---|---|---|---|
| Number of patients with any episode of culture-positive sepsis | 25 (1.5) | 68 (2.5) | 51 (1.9) | 0.038 [OR, 1.61 (95% CI: 1.02–2.56)] | 0.4 [OR, 1.23 (95% CI: 0.76–2.00)] |
| Number of episodes of culture-positive sepsis* | N = 25 | N = 68 | N = 51 | ||
| 1 episode | 23 (92) | 59 (86.8) | 49 (96.1) | 0.6 | 0.4 |
| 2 episodes | 2 (8) | 7 (10.3) | 2 (3.9) | ||
| 3 episodes | 0 (0) | 2 (2.9) | 0 (0) | ||
| Total number of episodes of culture-positive sepsis | 27 | 79 | 53 | … | … |
| Cumulative duration of hospital stay, person-days | 17,210 | 28,387 | 27,432 | … | … |
| Incidence rate of culture-positive sepsis (episodes/1000 patient-days) | 1.57 | 2.78 | 1.93 | 0.004 [IRR, 1.77 (95% CI: 1.13–2.85)] | 0.19 [IRR, 1.23 (95% CI: 0.76–2.04)] |
| Number of episodes with specific organisms† | N = 27 | N = 79 | N = 53 | ||
| E. coli | 4 (14.8) | 11 (13.9) | 9 (16.9) | 0.03 | 0.049 |
| K. pneumoniae | 3 (11.1) | 3 (3.8) | 4 (7.5) | ||
| A. baumannii | 0 (0) | 19 (24.0) | 11 (20.7) | ||
| S. aureus | 0 (0) | 5 (6.3) | 0 (0) | ||
| S. epidermidis | 4 (14.8) | 15 (18.9) | 14 (26.4) | ||
| Other CONS | 7 (25.9) | 16 (20.2) | 8 (15.1) | ||
| E fecalis/fecium | 1 (3.7) | 1 (1.3) | 0 (0) | ||
| Miscellaneous | 8 (29.6) | 9 (11.4) | 7 (13.2) | ||
| Incidence rate of A baumannii sepsis (episodes/1000 patient-days) | 0 | 0.67 | 0.4 | 0.0001 [IR difference, 0.67 (95% CI: 0.37–0.97)] | 0.0024 [IR difference, 0.40 (95% CI: 0.16–0.64)] |
| Incidence rate of non-A baumannii sepsis (episodes/1000 patient-days) | 1.57 | 2.11 | 1.53 | 0.09 [IRR, 1.35 (95% CI: 0.84–2.21)] | 0.46 [IRR, 0.98 (95% CI: 0.59–1.65)] |
| Antibiotic resistance among total number of episodes of sepsis† | N = 27 | N = 79 | N = 53 | ||
| Not MDR or worse | 15 (55.5) | 18 (22.8) | 12 (22.6) | 0.01 | 0.01 |
| MDR | 9 (33.3) | 35 (44.3) | 31 (58.5) | ||
| XDR | 3 (11.1) | 25 (31.6) | 10 (18.9) | ||
| PDR | 0 (0) | 1 (1.3) | 0 (0) | ||
| Incidence rate of MDR or worse (episodes/1000 patient-days) | 0.7 | 2.1 | 1.5 | <0.001 [IRR, 3.08 (95% CI: 1.64, 6.29)] | 0.007 [IRR, 2.14 (95% CI: 1.11, 4.48)] |
| Number of episodes of MDR or worse A. baumannii sepsis | 0 | 18 | 10 | 0.007 | 0.02 |
| (0†) | (22.8†) | (18.9†) | |||
| (94.7‡) | (90.1‡) | ||||
| Number of episodes of MDR or worse non–A. baumannii sepsis† | 12 (44.4) | 43 (54.4) | 31 (58.5) | 0.37 | 0.23 |
| Any culture-positive episode treated with meropenem* | 19 (76.0) | 59 (86.8) | 44 (86.3) | 0.2 [OR, 1.23 (95% CI: 0.47–3.23)] | 0.3 [OR, 2.04 (95% CI: 0.69–6.25)] |
| Any culture-positive episode treated with colistin* | 5 (20.0) | 32 (47.1) | 19 (37.3) | 0.018 [OR, 3.03 (95% CI: 1.03–8.33)] | 0.13 [OR, 2.44 (95% CI: 0.80–7.69)] |
| Patient care area where episodes of sepsis occurred† | N = 27 | N = 79 | N = 53 | ||
| Labor room stabilization | 3 (11.1) | 27 (34.2) | 18 (33.9) | 0.02 | 0.003 |
| NICU | 17 (62.9) | 45 (56.9) | 33 (62.3) | ||
| HDU | 7 (25.9) | 7 (8.8) | 2 (3.8) | ||
| Age 1st episode of sepsis, d, median (1st, 3rd quartile) | 5 (3, 13.5) | 5 (3, 8) | 6 (3, 9) | 0.7 | 0.7 |
| 1st episode of sepsis categorized by time of onset* | |||||
| EONS | 11 (44.0) | 25 (36.8) | 13 (25.5) | 0.5 [OR for LONS = 1.35 (95% CI: 0.53–3.43)] | 0.1 [OR for LONS = 2.30 (95% CI: 0.84–6.31)] |
| LONS | 14 (56.0) | 43 (63.2) | 38 (74.5) | ||
Percent expressed of number of patients with any episode of culture-positive sepsis.
Percent expressed of total number of episodes of culture-positive sepsis.
Percent expressed of all episodes of A. baumannii sepsis.
CONS indicates coagulase-negative Staphylococcus; EONS, early-onset neonatal sepsis; IR, incidence rate; IRR, IR ratio; LONS, late-onset neonatal sepsis; OR, odds ratio.
There were significantly more episodes caused by MDR/XDR/PDR organisms in group pre-LD and corres-LD compared with LD. The IRR of episodes caused by organisms that were MDR or worse when comparing group pre-LD versus LD [3.08 (95% CI: 1.64–6.29)] and group corres-LD versus LD [2.14 (95% CI: 1.11–4.48)] were statistically significant. Significantly more culture-positive episodes were treated with colistin in group pre-LD compared with group LD. Almost all isolates of A. baumannii were MDR/XDR/PDR. The difference in MDR/XDR/PDR organisms between the phases was primarily driven by the difference in A baumannii. There were no significant differences in the proportion of non–A. baumannii MDR organisms (as a percentage of all sepsis) between LD, pre-LD and corres-LD.
In group LD, 425 neonates were briefly isolated from their mothers and other neonates until their mothers’ COVID-19 report was available. One-hundred-eight severe acute respiratory syndrome coronavirus-2–positive mothers were isolated from other patients in a dedicated COVID building along with their neonates. Eight of these neonates were detected to be severe acute respiratory syndrome coronavirus-2 positive.
When comparing the phases of LD and unlocking, there were significant differences in gestational age and birth weight across phases, but no significant linear trends (Table 3). The proportion of subjects with any episode of culture-positive sepsis declined from LD phase 1.0 to unlock phase 4.0 but was not statistically significant. The difference in the proportion of etiologic organisms did not show a linear trend. There were progressively fewer episodes of MDR sepsis (Plinear trend = 0.047).
TABLE 3.
Comparison of the Phases of Lockdown and Unlocking
| Variables | Lockdown Phases 1 and 2 (N = 387) | Lockdown Phase 3 and 4 (N = 252) | Unlock Phase 1 (N = 310) | Unlock Phase 2 (N = 339) | Unlock Phase 3 (N = 296) | Unlock Phase 4 (N = 266) | P Value of Linear Trends* | P Value Without Linear Trends† |
|---|---|---|---|---|---|---|---|---|
| Number of live births per day, mean (SD) | 8.9 (2.8) | 9.0 (3.2) | 10.3 (3.2) | 10.9 (4.3) | 9.6 (3.3) | 8.8 (3.0) | 0.7 | 0.067 |
| Gestational age, weeks, median (1st, 3rd quartile) | 37 (35, 38) | 37 (34, 38) | 36 (34, 38) | 36 (34, 38) | 37 (34, 38) | 37 (34, 38) | 0.6 | 0.028 |
| Gestation categories‡ | ||||||||
| Extremely preterm (<28 weeks) | 8 (2.1) | 10 (4.0) | 10 (3.2) | 8 (2.4) | 7 (2.4) | 7 (2.7) | 0.3 | 0.059 |
| Very preterm (28–<32 weeks) | 17 (4.4) | 22 (8.7) | 28 (9.0) | 31 (9.1) | 23 (7.8) | 26 (10.0) | ||
| Moderate to late preterm§ | 130 (33.8) | 82 (32.5) | 120 (38.7) | 128 (37.8) | 104 (35.1) | 76 (29.1) | ||
| Term (37–41 weeks) | 230 (59.7) | 138 (54.8) | 152 (49.0) | 170 (50.1) | 162 (54.7) | 152 (58.2) | ||
| Post-term (>41 weeks) | 0 | 0 | 0 | 2 (0.6) | 0 | 0 | ||
| Birth weight, grams, median (1st, 3rd quartile) | 2521 | 2500 | 2325 | 2523 | 2512 | 2444 | 0.7 | 0.008 |
| (1975, 2897.5) | (1862.25, 2930) | (1748, 2721.5) | (1910, 2878) | (1907, 2918) | (1900.75, 2878) | |||
| Birth weight categories‡ | ||||||||
| <1000 g | 10 (2.6) | 12 (4.8) | 13 (4.2) | 14 (4.1) | 9 (3.0) | 12 (4.6) | 0.3 | 0.056 |
| 1000–1499 g | 22 (5.7) | 31 (12.3) | 37 (11.9) | 32 (9.4) | 26 (8.8) | 27 (10.3) | ||
| 1500–2499 g | 153 (39.7) | 79 (31.3) | 133 (42.9) | 114 (33.6) | 106 (35.8) | 99 (37.9) | ||
| 2500–4200 g | 199 (51.7) | 130 (51.6) | 127 (41.0) | 179 (52.8) | 155 (52.4) | 123 (47.1) | ||
| >4200 g | 1 (0.3) | 0 | 0 | 0 | 0 | 0 | ||
| Sex‡ | ||||||||
| Male | 213 (55.5) | 141 (56.0) | 178 (57.4) | 178 (52.5) | 151 (51.0) | 149 (56.0) | 0.7 | 0.4 |
| Female | 174 (45.0) | 111 (44.0) | 132 (42.6) | 159 (46.9) | 145 (49.0) | 116 (43.6) | ||
| Ambiguous | 0 (0) | 0 (0) | 0 (0) | 2 (0.6) | 0 (0) | 1 (0.4) | ||
| Appropriateness for gestation‡ | ||||||||
| AGA | 221 (61.2) | 169 (68.1) | 185 (63.4) | 210 (64.8) | 194 (67.6) | 157 (60.6) | 0.9 | 0.4 |
| SGA | 130 (36.0) | 77 (31.0) | 99 (33.9) | 107 (33.0) | 84 (29.3) | 97 (37.5) | ||
| LGA | 10 (2.8) | 2 (0.8) | 8 (2.7) | 7 (2.2) | 9 (3.1) | 5 (1.9) | ||
| Apgar score at 5 min | 8 (7, 8) | 8 (7, 8) | 8 (7, 9) | 8 (7, 8) | 8 (7, 9) | 8 (7, 8) | 0.04 | 0.02 |
| Admitted in NICU‡ | 36 (9.3) | 26 (10.3) | 37 (11.9) | 39 (11.5) | 34 (11.5) | 13 (11.3) | 0.3 | 0.9 |
| Admitted in HDU‡ | 80 (20.7) | 50 (19.8) | 54 (17.4) | 46 (13.6) | 62 (20.9) | 24 (9.0) | 0.002 | <0.001 |
| Age transferred to NICU, d, median (1st, 3rd quartile) | 2 (2, 2.75) | 2 (1, 2.25) | 2 (1.5, 3) | 2 (1, 3) | 3 (1, 5) | 2 (1, 4.25) | 0.13 | 0.4 |
| Age transferred to HDU, d, median (1st, 3rd quartile) | 4 (2, 8) | 7 (3, 15) | 9 (5, 12.25) | 8 (5, 13) | 7 (4, 15.25) | 5.5 (2.25, 10) | 0.018 | 0.001 |
| Any episode of culture- positive sepsis‡ | 8 (2.1) | 4 (1.6) | 3 (1.0) | 6 (1.8) | 4 (1.4) | 3 (1.1) | 0.4 | 0.9 |
| Total number of episodes with specific organisms | N = 9 | N = 4 | N = 4 | N = 6 | N = 4 | N = 5 | ||
| E. coli | 1 | 1 | 2 | 0 | 0 | 1 | 0.1 | 0.04 |
| K. pneumoniae | 0 | 0 | 0 | 1 | 2 | 0 | ||
| A. baumannii | 0 | 0 | 0 | 0 | 0 | 2 | ||
| P. aeruginosa | 0 | 0 | 0 | 1 | 1 | 0 | ||
| S. epidermidis | 2 | 0 | 0 | 2 | 0 | 0 | ||
| Other CONS | 2 | 3 | 1 | 1 | 0 | 1 | ||
| E fecalis/fecium | 2 | 0 | 0 | 0 | 0 | 0 | ||
| Miscellaneous | 2 | 0 | 1 | 1 | 1 | 1 | ||
| Antibiotic resistance among total number of episodes of sepsis | N = 9 | N = 4 | N = 4 | N = 6 | N = 4 | N = 5 | ||
| Not MDR or worse | 2 | 3 | 3 | 3 | 4 | 1 | 0.047 | 0.9 |
| MDR | 4 | 1 | 1 | 3 | 0 | 1 | ||
| XDR | 3 | 0 | 0 | 0 | 0 | 3 | ||
| Culture-positive episode treated with meropenem¶ | 5 (62.5) | 3 (75) | 3 (100) | 4 (66.7) | 4 (100) | 3 (100) | 0.1 | 0.5 |
| Culture-positive episode treated with colistin¶ | 1 (12.5) | 0 (0) | 2 (66.7) | 2 (33.3) | 0 (0) | 1 (33.3) | 0.5 | 0.2 |
| Outcome‡ | ||||||||
| Discharged home | 370 (95.6) | 241 (95.6) | 297 (95.8) | 323 (95.3) | 285 (96.3) | 252 (94.7) | 0.6 | 0.7 |
| Died | 17 (4.4) | 10 (4.0) | 13 (4.2) | 15 (4.4) | 11 (3.7) | 13 (4.9) | ||
| LAMA or DOR | 0 (0) | 1 (0.4) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | ||
| Transferred elsewhere | 0 (0) | 0 (0) | 0 (0) | 1 (0.3) | 0 (0) | 1 (0.4) | ||
| Duration hospital stay, d, median (1st, 3rd quartile) | 4 (3, 7) | 5 (3, 7) | 5 (4, 7) | 4 (4, 7) | 4 (3, 8) | 4 (3, 7) | 0.07 | 0.018 |
| Duration of hospital stay among <32 weeks gestation, d, median (1st, 3rd quartile) | 29 (10.5, 54.5) | 18.5 (5, 41) | 17.5 (2.75, 31.5) | 20 (4, 29) | 22 (3, 39) | 29 (8, 57) | 0.1 | 0.8 |
2 test for linear trends of outcome categorical variables. Jonckheere-Terpstra test for linear trends of outcome numerical variables.
2 test for outcome categorical variables. Kruskal-Wallis test for outcome numerical variables.
Percent expressed of total number in respective lockdown/unlock phase.
32–<37 weeks.
Percent expressed out of total number of culture-positive episodes in respective lockdown/unlock phase.
AGA indicates appropriate for gestational age; CONS, coagulase-negative Staphylococcus; DOR, discharged on request; LAMA, left against medical advice; LGA, large for gestational age; SGA, small for gestational age.
The candidate predictors in the multivariable analysis did not show multi-collinearity, hence all were included in the model. The Group (ie, LD, pre-LD and corres-LD), male sex, birth weight and Apgar score independently predicted the occurrence of any episode of culture-positive sepsis (Table 4). The prediction accuracy was 97.9% (assuming cases with >50% predicted probability were true sepsis) and the C-statistic was 0.85 (95% CI: 0.81–0.89).
TABLE 4.
Multivariable Logistic Regression Model to Predict Patients With Any Episode of Culture-Positive Sepsis
| Variable | Regression Coefficient | aOR | 95% CI of aOR | P value |
|---|---|---|---|---|
| Constant | −2.25 | |||
| Time period | 0.053 | |||
| LD (reference group) | ||||
| Pre-LD | 0.81 | 2.24 | 1.11–4.51 | 0.024 |
| Corres-LD | 0.51 | 1.67 | 0.83–3.35 | 0.15 |
| Births in the week subject was born | −0.007 | 0.99 | 0.98–1.004 | 0.2 |
| Male sex | 0.81 | 2.26 | 1.55–3.29 | <0.001 |
| Gestational age | −0.05 | 0.95 | 0.48–1.08 | 0.5 |
| Birth weight | −0.002 | 0.998 | 0.997–0.999 | <0.001 |
| Apgar score at 5 min | 0.53 | 1.69 | 1.46–1.96 | <0.001 |
| Appropriateness for gestation | 0.2 | |||
| AGA (reference group) | ||||
| SGA | −0.4 | 0.67 | 0.39–1.16 | 0.15 |
| LGA | 0.92 | 2.5 | 0.64–9.73 | 0.19 |
AGA indicates appropriate for gestational age; aOR, adjusted odds ratio; LGA, large for gestational age; SGA, small for gestational age.
To better understand A. baumannii, which had disappeared in group LD, we pooled data of the first episode of Gram-negative sepsis from groups pre-LD and corres-LD and compared A. baumannii versus other Gram-negative sepsis (see Table, Supplementary Digital Content 2, http://links.lww.com/INF/E681). On univariable analysis, A baumannii sepsis was associated with male sex, lower gestational age, lower birth weight and more concurrent births in the week the subject was born. In a multivariable logistic regression analysis to predict A baumannii sepsis among all cases with Gram-negative sepsis in the full study population (see Table, Supplementary Digital Content 3, http://links.lww.com/INF/E682), the only independent predictor was the number of concurrent births.
DISCUSSION
The number of live births in group LD was only 60% of the numbers in group pre-LD and corres-LD. The reduced number of live births in group LD was probably due to difficulties encountered by patients in traveling to our hospital during the LD and the fear of contracting COVID-19 at our hospital. Many mothers may have preferred to deliver in a hospital closer to home. Neonates in group LD had better Apgar scores and were discharged with better outcomes. Greater proportions in group LD were admitted in the NICU and HDU, and at a younger postnatal age. There was an overall reduction in culture-proven sepsis following the LD primarily driven by a striking reduction in A. baumannii sepsis and sepsis due to MDR/XDR/PDR organisms. This led to lower usage of colistin. Patients in the pre-LD group (compared with group LD), male sex, lower birth weight and Apgar score were independent predictors of any episode of culture-positive sepsis. It was reassuring to find that there was no increase in the incidence of culture-proven sepsis with progressive unlocking.
There were no clinically meaningful differences in the average gestation between the three groups, although there was a statistically significant difference between LD and corres-LD. Our observation is contrary to previous reports, where the incidence of preterm births has declined during the pandemic.5,17 More neonates were admitted to the NICU and HDU after the LD possibly because the decrease in absolute number of live births made it easier to find a vacant bed in our NICU or HDU. The lower postnatal age of transfer to these areas supports the latter possibility. Similarly, the relatively longer hospital stay in the LD group can be attributed to the greater availability of vacant beds and hence less pressure on the physicians to discharge patients before schedule from the hospital.
The decrease in hospital deliveries coupled with an increase in the need for NICU/HDU care has not been reported before. However, a 57% decrease in children’s emergency visits has been documented during the peak of the pandemic.18–21 Like us, Goldman et al found that the acuity of sickness and admission rate among children was higher during the peak compared with preceding epochs. In a recent meta-analysis, Chmielewska et al did not observe significant differences in the mode of delivery, births <32 weeks gestation, perinatal asphyxia, low-birth-weight status, NICU admissions or neonatal deaths between the pandemic and pre-pandemic periods.5 The incidence of preterm births <37 weeks gestation was significantly lower only among high-income countries. This meta-analysis did not evaluate neonatal sepsis.
The improvement in the epidemiology of neonatal sepsis in our study may be partly explained by earlier transfer of neonates to the NICU and HDU than what was possible prior to the LD. This is reflected in fewer episodes of sepsis in the labor room stabilization unit. The changes can also be attributed to behavioral factors. Unfortunately, we do not have data that can quantify these behavioral changes, so our observations are necessarily anecdotal and subjective. Prior to the LD, healthcare workers (HCW) and family members were expected to wash hands at entry to all patient care areas and HCWs were expected to follow hand hygiene measures at the 5 moments of hand hygiene prescribed by the World Health Organization but wearing a face mask was not mandatory. During the LD, compliance to hand hygiene measures increased substantially and there was universal usage of face masks by HCWs and family members. Before the LD, housekeeping measures included regular disinfection of equipment and surfaces, but during the LD these were followed more assiduously, with disinfection of high-contact surfaces such as door handles, telephones, etc. performed at least twice in every nursing shift. Instructions for disinfection of mobile phones were provided to all HCWs. Patients were distributed across more areas because those awaiting COVID reports were isolated in a separate 12-bedded isolation ward for mothers and a 6-bedded isolation ward for neonates, and proven cases were managed in a separate building. HCWs working in the isolation wards with suspected or proven cases of COVID-19 donned N-95 masks, caps, goggles, impervious gowns, shoe covers and gloves. There were no changes in the water sources or frequency of surveillance cultures for MDR organisms in the unit. Academic activities and meetings went online, and HCWs practiced social distancing.
A. baumannii is an important cause of neonatal sepsis in developing countries.22 The decrease in A. baumannii sepsis disproportionate to other Gram-negative sepsis in group LD is puzzling. The only independent risk factor we could identify was less overcrowding (see Table, Supplemental Digital Content 4, http://links.lww.com/INF/E683), but it is unclear why it should preferentially impact one species. A. baumannii is known to behave differently than other Gram-negative bacteria, characterized by more rapid acquisition of MDR, and resistance to desiccation and disinfectants.23 Compared with E. coli or Klebsiella, neonates with A. baumannii sepsis have lower birth weight.24 It is possible that infection control measures may have differential effects on A. baumannii sepsis and other Gram-negative sepsis. In a systematic review and network meta-analysis of 42 studies on adult ICU patients, compared with standard treatment alone, the addition of environmental cleaning to standard treatment plus antibiotic stewardship program or the addition of source control to standard treatment and environmental cleaning significantly reduced MDR A. baumannii sepsis.25 On the other hand, infection control strategies that mandatorily included antibiotic stewardship programs significantly reduced sepsis due to MDR extended-spectrum beta-lactamase producing Gram-negative infections. In a study from China, the authors reported the effect of relocation of the NICU and introduction of a care bundle for prevention of ventilator-acquired pneumonia (VAP).26 The incidence of VAP due to A. baumannii decreased from 79% to 36%, whereas VAP due to K. pneumoniae increased from 17% to 18%. Infection control measures rapidly control outbreaks of A. baumannii in neonatal units.27,28 Our study was not specifically designed to evaluate the effect of infection control measures on the decline of MDR sepsis or organism-wise sepsis and we were unable to quantify the infection control measures. Although augmentation in infection control measures was an important feature of the LD epoch, apart from infection control measures there were several other differences between the 3 epochs. Authors of a study on older patients in the United Kingdom reported differential effects of the pandemic on etiologic organisms: Enterobacteriaceae reached an all-time low and coagulase-negative Staphylococcus an all-time high.29
The absolute decline in MDR sepsis was not merely a consequence of overall decline in sepsis, as MDR sepsis as a proportion of all sepsis declined significantly in group LD. Our data shows that the decrease in MDR sepsis was driven by the decline in A. baumannii, which is frequently multi-drug resistant.
As early-onset neonatal sepsis is attributed to maternal risk factors that are unlikely to be altered by the pandemic, and late-onset neonatal sepsis is environmentally acquired, we expected to see a greater proportion of early-onset neonatal sepsis in group LD. We observed this difference, but it was not statistically significant.
It was interesting to note that there were no major differences from LD 1.0 to unlock 4.0. This could be due to the small sample sizes in each phase. The easing of restrictions may have had little impact on hospital functioning. It did not alter important predictors of sepsis, for example, gestational age, birth weight and appropriateness for gestational age. A closer look at the government stringency index shows that there was very little objective decline in the stringency of restrictions during the various phases of unlocking. We do not have data to quantify any change in behavior or infection control measures during the phases of unlocking.
In the multivariable analysis, Group pre-LD independently doubled the odds of an episode of culture-positive sepsis compared with group LD. Our analysis suggests that the overall changes after the LD were associated with less culture-positive sepsis rather than overcrowding specifically around the time of birth.
Groups LD and corres-LD covered virtually identical dates exactly a year apart whereas group pre-LD covered a different time of the year. Since the differences in most outcomes were larger between groups LD and pre-LD compared with groups LD and corres-LD, it suggests that seasonal differences may have accentuated the differences between Groups LD and pre-LD. We studied the incidence of sepsis, particularly A. baumannii sepsis, for 4 years prior to 2019, to evaluate the possibility that a downward sloping trend may have exaggerated the effect of the LD. However, no such trend could be observed from the year 2015 through 2018 (see Table, Supplemental Digital Content 4, http://links.lww.com/INF/E683).
The strengths of our study were a large sample size and two comparison groups. Being retrospective, it was prone to several biases. There may have been unmeasured differences in patient characteristics, clinical practices, and rigor of data collection between the time periods. We limited our observation to hospital stay and may have missed cases of sepsis after transfer or discharge. Our results may not be generalizable to centers with fewer resource constraints, and less MDR sepsis. Weekly births are a sub-optimal surrogate for overcrowding. We were unable to quantify behavioral changes and infection control measures following the LD and during unlocking. We do not have data about the harm caused to neonates whose mothers were not able to reach our center due to the pandemic. Improvement in sepsis parameters may have occurred at the expense of worse outcomes among those who delivered elsewhere.
CONCLUSIONS
From the limited experience accrued from our center, we conclude that the LD may be associated with a decrease in culture-proven neonatal sepsis, primarily due to A. baumannii sepsis and antibiotic-resistant sepsis in settings like ours. Efforts must be made to consolidate the gains achieved during the pandemic by reinforcing the desirable administrative and infection control practices.
Supplementary Material
Footnotes
The authors have no conflicts of interest or sources of funding relevant to this article to disclose.
Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s website (www.pidj.com)
References
- 1.Worldometer. Available at: https://www.worldometers.info/coronavirus/ Accessed April 7, 2021.
- 2.Burki T. The indirect impact of COVID-19 on women. Lancet Infect Dis. 2020;20:904–905. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Masroor S. Collateral damage of COVID-19 pandemic: delayed medical care. J Card Surg. 2020;35:1345–1347. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Roberton T, Carter ED, Chou VB, et al. Early estimates of the indirect effects of the COVID-19 pandemic on maternal and child mortality in low-income and middle-income countries: a modelling study. Lancet Glob Health. 2020;8:e901–e908. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Chmielewska B, Barratt I, Townsend R, et al. Effects of the COVID-19 pandemic on maternal and perinatal outcomes: a systematic review and meta-analysis. Lancet Glob Health. 2021;9:e759–e772. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Sakamoto H, Ishikane M, Ueda P. Seasonal influenza activity during the SARS-CoV-2 outbreak in Japan. JAMA. 2020;323:1969–1971. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Partridge E, McCleery E, Cheema R, et al. Evaluation of seasonal respiratory virus activity before and after the statewide COVID-19 shelter-in-place order in Northern California. JAMA Netw Open. 2021;4:e2035281. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Angoulvant F, Ouldali N, Yang DD, et al. Coronavirus disease 2019 pandemic: impact caused by school closure and national lockdown on pediatric visits and admissions for viral and nonviral infections-a time series analysis. Clin Infect Dis. 2021;72:319–322. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Kuitunen I, Artama M, Mäkelä L, et al. Effect of social distancing due to the COVID-19 pandemic on the incidence of viral respiratory tract infections in children in finland during early 2020. Pediatr Infect Dis J. 2020;39:e423–e427. [DOI] [PubMed] [Google Scholar]
- 10.Lee H, Lee H, Song KH, et al. Impact of public health interventions on seasonal influenza activity during the COVID-19 outbreak in Korea. Clin Infect Dis. 2021;73:e132–e140. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Nolen LD, Seeman S, Bruden D, et al. Impact of social distancing and travel restrictions on Non-Coronavirus Disease 2019 (Non-COVID-19) respiratory hospital admissions in young children in Rural Alaska. Clin Infect Dis. 2021;72:2196–2198. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Wee LE, Conceicao EP, Sim XYJ, et al. Reduction in healthcare-associated respiratory viral infections during a COVID-19 outbreak. Clin Microbiol Infect. 2020;26:1579–1581. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Chaurasia S, Sivanandan S, Agarwal R, et al. Neonatal sepsis in South Asia: huge burden and spiralling antimicrobial resistance. BMJ. 2019;364:k5314. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Covid-19 government response tracker. Available at: https://www.bsg.ox.ac.uk/research/research-projects/covid-19-government-response-tracker. Accessed October 10, 2020.
- 15.Covid-19 lockdown in India. Available at: https://en.wikipedia.org/wiki/COVID-19_lockdown_in_India#Prohibitions2021. Accessed March 3, 2021.
- 16.Magiorakos AP, Srinivasan A, Carey RB, et al. Multidrug-resistant, extensively drug-resistant and pandrug-resistant bacteria: an international expert proposal for interim standard definitions for acquired resistance. Clin Microbiol Infect. 2012;18:268–281. [DOI] [PubMed] [Google Scholar]
- 17.Lemon L, Edwards RP, Simhan HN. What is driving the decreased incidence of preterm birth during the coronavirus disease 2019 pandemic? Am J Obstet Gynecol MFM. 2021;3:100330. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Hughes HE, Hughes TC, Morbey R, et al. Emergency department use during COVID-19 as described by syndromic surveillance. Emerg Med J. 2020;37:600–604. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Lazzerini M, Barbi E, Apicella A, et al. Delayed access or provision of care in Italy resulting from fear of COVID-19. Lancet Child Adolesc Health. 2020;4:e10–e11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Nuñez JH, Sallent A, Lakhani K, et al. Impact of the COVID-19 pandemic on an emergency traumatology service: experience at a Tertiary Trauma Centre in Spain. Injury. 2020;51:1414–1418. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Goldman RD, Grafstein E, Barclay N, et al. Paediatric patients seen in 18 emergency departments during the COVID-19 pandemic. Emerg Med J. 2020;37:773–777. [DOI] [PubMed] [Google Scholar]
- 22.Investigators of the Delhi Neonatal Infection Study c. Characterisation and antimicrobial resistance of sepsis pathogens in neonates born in tertiary care centres in Delhi, India: a cohort study. Lancet Glob Health. 2016;4:e752–e760. [DOI] [PubMed] [Google Scholar]
- 23.Peleg AY, Seifert H, Paterson DL. Acinetobacter baumannii: emergence of a successful pathogen. Clin Microbiol Rev. 2008;21:538–582. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Hsu JF, Chu SM, Lien R, et al. Case-control analysis of endemic Acinetobacter baumannii bacteremia in the neonatal intensive care unit. Am J Infect Control. 2014;42:23–27. [DOI] [PubMed] [Google Scholar]
- 25.Teerawattanapong N, Kengkla K, Dilokthornsakul P, et al. Prevention and control of multidrug-resistant gram-negative bacteria in adult intensive care units: a systematic review and network meta-analysis. Clin Infect Dis. 2017;64(suppl_2):S51–S60. [DOI] [PubMed] [Google Scholar]
- 26.Zhou Q, Lee SK, Jiang SY, et al. Efficacy of an infection control program in reducing ventilator-associated pneumonia in a Chinese neonatal intensive care unit. Am J Infect Control. 2013;41:1059–1064. [DOI] [PubMed] [Google Scholar]
- 27.Melamed R, Greenberg D, Porat N, et al. Successful control of an Acinetobacter baumannii outbreak in a neonatal intensive care unit. J Hosp Infect. 2003;53:31–38. [DOI] [PubMed] [Google Scholar]
- 28.Tsiatsiou O, Iosifidis Ε, Katragkou A, et al. Successful management of an outbreak due to carbapenem-resistant Acinetobacter baumannii in a neonatal intensive care unit. Eur J Pediatr. 2015;174:65–74. [DOI] [PubMed] [Google Scholar]
- 29.Denny S, Rawson TM, Hart P, et al. Bacteraemia variation during the COVID-19 pandemic; a multi-centre UK secondary care ecological analysis. BMC Infect Dis. 2021;21:556–564. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.