This observational and analytical study of diagnostic test outcome was carried out over a period of 12 months from January-December 2016 in Rajarshi Chhatrapati Shahu Maharaj Medical College and Chhatrapati Pramilaraje Rugnalaya, Kolhapur, Maharashtra, India. Institutional Ethical Committee approval was obtained (IEC no. ECR/703/Inst/MH/2015/RR-20). Consecutive cases of 104 neonates with neonatal sepsis which were prospectively registered during the period from January - December 2016 were enrolled in the study. Informed consent was taken from the parents of the neonates.

Inclusion criteria: The neonates admitted in neonatal intensive care unit with clinical suspicion of sepsis were included in the study.

Exclusion criteria: The neonates with hyaline membrane disease, transient tachypnea of newborn or hypoxia induced encephalopathy were excluded from the study.

The study population was evaluated thoroughly for clinical course, gestational age, birth weight and age of onset of sepsis. The details of maternal conditions were noted. Maternal and foetal risk factors were studied diligently.

Under all aseptic precautions 5 mL of intravenous blood was collected with a disposable 5 mL syringe. It was distributed as follows, 1 mL each in Ethylenediamine Tetra-acetic Acid (EDTA) and plain microtainer and 3 mL in brain heart infusion broth for following investigations:

Based on blood culture positivity, cases were classified as proven and probable neonatal sepsis and were compared with each other for calculating diagnostic utility. The positive blood culture was considered as gold standard for diagnosis of sepsis.

Following is the list of haematological parameters studied with mention of values decided as abnormal.

Haemoglobin (Hb) <14 g/dL, Total Leucocyte Count (TLC) <5000/cmm or >20,000/cmm, ANC <1800/cmm, I:T ratio (I:T) ≥0.2, Platelet Count (PC) <1,50,000/cmm, Mean Platelet Volume (MPV) >11.5 fL, Platelet Distribution Width (PDW) > 13.5% and presence of TG on peripheral blood smear [5-9].

The data was analysed to find out the clinical profile of neonatal sepsis, its risk factors, outcome profile, haematological profile, and diagnostic utility of various parameters including CRP and haematological parameters in isolation and in various combinations. Here combinations were studied with “OR” values. The diagnostic utility of conventional septic screen score with combination of either pure haematological parameters (TLC, ANC, I:T ratio, TG, PC) or those with CRP was carried out with “AND” values. The score was considered satisfactory when three or more criteria were fulfilled.

Sensitivity, Specificity, PPV, NPV and DA were calculated with appropriate formulae. For calculation of diagnostic utility of various combinations of test “Parallel” method was used with "AND" and “OR” values. The p-value was calculated by Chi-square test to study frequency of occurrence.

There were 104 cases of neonatal sepsis of with 61 (58.7%) males and 43 (41.3%) females with Male:Female ratio as 1.4:1. In 73 (70.2%) cases blood culture was positive. These cases were classified as proven sepsis. While 31 (29.8%) cases did not yield organisms on culture and were classified as probable sepsis.

Clinical profile of neonatal sepsis was as follows: The early onset sepsis which presented within three days of birth was seen in 67 (64.4%) neonates while late onset sepsis which presented between 4-30 days was seen in 37 (35.6%) neonates. Early onset sepsis was statistically more common than late onset sepsis with p-value=0.066. The sepsis was more common in low birth weight neonates with weight less than 2500 grams (n=84, 80.8%) than in those with normal birth weight (more than or equal to 2500 grams) with p-value=0.028 (n=20, 19.2%). A higher frequency of sepsis was noted in preterm neonates (n=72, 69.2%) as compared to full term neonates (n=32, 30.8%) with p-value=0.011. There was no statistically significant influence of type of delivery on occurrence of sepsis (p-value=0.72) with 43 (41.3%) neonates being delivered as by caesarean section as against 61 (58.7%) being delivered with normal vaginal route.

The maternal risk factors for neonatal sepsis were premature rupture of membranes, pregnancy induced hypertension, prolonged labour, both these complications seen concurrently with number of cases being-10 (9.61%), 8 (7.69%), 7 (6.73%) and 6 (5.76%), respectively. Foetal factors consisted of low birth weight, prematurity, meconium aspiration, twin pregnancy and prolonged hospitalisation (n=84 (80.76%), 72 (69.23%), 9 (8.65%), 4 (3.84%) and 2 (1.92%) cases, respectively.

The culture proven sepsis was significantly more associated with low birth weight with p-value=0.02. The early onset sepsis was significantly more associated with prematurity (p-value=0.01).

There occurred 32 fatalities (30.8%), 27 of these were associated with preterm birth. Thirty deaths were seen in neonates with low birth weight. The blood culture positivity was a feature of 29 fatal cases. Preterm birth, low birth weight and positive blood culture were significantly associated with mortality with p-value=0.02. Type of delivery, gender of neonate and time of onset of sepsis did not affect the fatality significantly with p-values of 0.74, 0.92 and 0.54. The abnormalities of sepsis screen parameters that mattered in fatal cases were raised I:T ratio and thrombocytopenia with p-value <0.05.

Of the 73 culture positive cases, gram negative organisms were grown in 29 (39.7%) cases, the gram positive organisms were seen in 21 (28.8%) cases and Candida was noted in 23 (31.5%) cases. The gram negative organisms included Klebsiella pneumoniae (n:13), E. coli (n=6) and Acinetobacter (n=6) and Pseudomonas aeruginosa (n=4). The gram positive organisms consisted of coagulase negative Staphylococcus (n=4), methicillin resistant coagulase negative staphylococci (n=10), Staphylococcus aureus (n=6),) and Streptococci (n=1).

Haematological and CRP profile of cases of neonatal sepsis is summarised as [Table/Fig-1].

In proven sepsis, there was one case which only showed raised CRP and succumbed to death. There were 33 cases which showed three or more conventional haematological parameters of sepsis to be abnormal. Forty two cases fulfilled three or more sepsis screen criteria with combinations with CRP with overlap of cases. There occurred seven deaths in these cases. In probable sepsis six cases fulfilled three or more sepsis screen haematological criteria and six more cases showed three point sepsis screen positivity when CRP was considered. One case having only thrombocytopenia as an abnormality had a mortality. This particular neonate was born of twin delivery. In three cases haematological parameters and CRP was normal.

The diagnostic utility of individual haematological parameters and CRP in neonatal sepsis is depicted in [Table/Fig-2].

It can be seen from [Table/Fig-2] that CRP showed highest sensitivity and haemoglobin had highest specificity. The overall diagnostic utility with highest PPV, NPV and DA was exhibited by I:T ratio.

The diagnostic utility of various combinations of conventional pure haematological septic screen parameters or those including CRP; necessitating positivity of three or more such parameters for diagnosis of culture proven neonatal sepsis was calculated. The results are shown in [Table/Fig-3].

It can be seen from [Table/Fig-3] that specificity and PPV of sepsis screen parameters except for combinations involving CRP and I:T ratio was in the range of 87.1-100% and 66.7-100%. The overall sensitivity, NPV and DA of the whole group was not very satisfactory. The frequency of occurrence of CRP with combinations with single or multiple parameters of sepsis showed significant difference in proven and probable sepsis group (p-value <0.05) except for combination of CRP+TLC+I:T ratio. Similarly pure combinations of haematological sepsis screen were studied for difference in frequency of occurrence. It was significant with p-value <0.05 in combinations of TLC+TG+PC, TLC+ANC+PC and combination of any three or more haematological parameters of sepsis screen.

Diagnostic utility of combinations of CRP with haematological parameters with OR values is seen as [Table/Fig-4].

The best 3 point combinations with CRP had TLC, I:T ratio or PC in common. Almost similar utility was shown by combination of CRP with all analyser parameters. This was surpassed only by combination of CRP with all haematological parameters. The utility of combination of pure haematological parameters is noted in following table [Table/Fig-5]. The best combinations of pure haematological parameters in terms of diagnostic utility were those including WBC parameters like TLC, I:T ratio and TG; only to be surpassed by combination of all haematological parameters.

Much is being said about of haematological parameters and CRP as a biomarker of neonatal sepsis in terms of their utility, accessibility, simplicity of estimation, cost effectiveness and ability to provide early diagnosis [3,10].

In a study by Rudwell RL et al., described haematological scoring system consisting of seven haematological parameters of neonatal sepsis [5]. These were TLC, ANL, I:T ratio, PC and severe degenerative changes in neutrophils and micro Erythrocyte Sedimentation Rate (ESR). Abnormalities in each of these parameters were scored as 1. The likely hood of sepsis was 31% when the scores were equal to or more than 3 and with score less than 3 the possibility of sepsis could be completely ruled out. Other studies used this scoring system in its pure form or its modification in assessing neonatal sepsis, some which augmented it with CRP [2,6,11-14].

The conventional haematological sepsis screen criteria of three or more parameters for diagnosis of sepsis yielded DA of 55.7% and with CRP it rose to 60.6% in present study. This seems too modest to be a result of an exercise. Interestingly, there were eight mortalities in cases showing less than 3 criteria of neonatal sepsis, two of these cases showed single abnormality in the form of raised CRP and thrombocytopenia in one case each. This implies that an exercise of using septic screen score with rigid criteria necessitating abnormalities of multiple parameters to be present to diagnose sepsis does not seem to be suitable for confirming cases of neonatal sepsis.

It can be inferred from present study that the best diagnostic utility was offered by combinations including CRP. A very liberal combination of all parameters including CRP could effectively pick up all cases of culture positive sepsis with best utility parameters. The diagnostic utility of pure combination of haematological parameters was also remarkable with 99% DA. This was followed by wholesome combination of WBC parameters.

Particularly impressive was the utility of wholesome combination of those parameters which can be exclusively obtained by 3 part automated haematology analyser with DA of 92.3%, PPV and specificity 100% and sensitivity of 89%.

The liberal use of combination of parameters with “OR” values was meant for finding out ability of these biomarkers in acting at par or to match the performance of blood culture to prove the sepsis. The magnitude to which it can be done is reflected in their diagnostic utility. Interestingly fair number of cases of probable sepsis with negative blood culture showed abnormal results for one or more of conventional biomarkers of sepsis (n=27, 87%).

Four of such cases succumbed to death. This implies that clinical suspicion of sepsis was valid at least in 87% of these cases though it cannot be corroborated with blood culture. If we duly considered two cases with raised PDW to be indicative of sepsis almost 93.5% cases of probable sepsis group confirmed clinical suspicion of neonatal sepsis.

The strength of present study lies in its very elaborate and comprehensive assessment of haematological biomarkers both with septic screen score, in their pure forms, in various liberal combinations including those with CRP in finding out their diagnostic utility. The cost effectiveness and accessibility of this exercise is also worth considering.

The major drawback of present study was that it did not involved serial estimations of parameters which would have enlightened us on response to antibiotic treatment. These parameters were not tested in normal age matched neonates on ethical grounds.

The combinations of various haematological parameters as obtained by analyser when used liberally increase the diagnostic utility and can act as alternative to sepsis scoring systems. It can obviate the need for estimation of CRP and further need of peripheral blood smear which needs expertise of pathologist. The almost ubiquitous availability of 3 part analysers makes such an attempt very accessible.