The incidence of hospital admission due to leptospirosis was significantly higher than that of non-leptospirosis infection admission [17]. Identifying patients who are at higher risk for prolonged hospitalization due to the disease will help reducing its burden in form of the coast of longer hospital stay. Recognizing these factors will help not only local practitioners, but also practitioners in other countries where the disease is not endemic but is transmitted through the travelers and emigrants where the awareness of health care providers regarding the disease burden including the financial burden is critically needed [32, 33]. This is more important nowadays than ever before as the incidence of the disease is expected to increase as a result of the global warming and increasing the international travelling and tourism [5, 34].
The average hospital stays varied in different studies and ranged from less than 5 days [35] to more than 10 days [25]. Our finding of mean LOS of 6.77 days is within the international range for the hospital stay in leptospirosis patients. Moreover, the mean LOS for patients with prolonged hospitalization in our study (13.3±7.7 days) was slightly longer comparing to another study (12±9 days) [36], but the mean LOS of patients without prolonged hospitalization in our study (4.46±1.53 days) was near to the LOS in another study (4.34±1.96 days) [37]. Moreover, our results showed more than quarter of the sample investigated (25.9%) had prolonged hospitalization in form of spending more than seven days and was not uncommon to find severe leptospirosis patients spent significantly longer LOS than mild form in a similar manner to the finding shown by another study [37]. However, 15.4% of patients with prolonged hospitalization had mild form and around half of patients (48.3%) without prolonged hospitalization had severe form in the current study.
Our study showed older patients (age > 40 years) significantly spent more than 7 days comparing to younger patients. However, male patients insignificantly had more prolonged hospitalization than females. Similarly, one study showed leptospirosis patients of male gender with age > 20 years were insignificantly more and had longer hospitalization, [17]. Moreover, another study showed that males had significantly longer hospital stay compared to females and individuals aged ≥ 38 had significantly longer hospital stay [38]. Similarly, LOPES, COSTA [39] found that older survivors significantly spent longer hospital stay than younger survivors. Furthermore, patients from rural areas significantly spent longer hospitalization period than their counterparts from urban area in the current study. This may be attributed to the longer period they spent before they were admitted to hospital. This delay in seeking care may be due to logistical barriers or deficiency in the awareness of those people regarding the disease. At the end, this delay is expected to increase the chance for further complications of the disease which necessitate further and longer period of hospital care for those patients. Mean value of the duration before the admission is significantly longer in patients who had prolonged hospitalization according to data from the current study, and late hospitalization after four days was insignificantly associated with the prolonged hospitalization. These findings underscore the need for awareness campaign particularly in the rural areas where the disease is mainly endemic to improve the care seeking behavior for those people and to help them access the hospital care as early as possible [40].
Surprisingly, neither smoking nor any of the studied comorbidities was significantly associated with the prolonged hospitalization, however all of the comorbidities but CHF (which was insignificantly more frequent in patients without prolonged hospitalization) were insignificantly more frequently encountered in patients with prolonged hospitalization. Moreover, even presence of more than one comorbidity was more frequent in the patients with prolonged hospitalization but did not reach the significant level (p=0.088). This may be attributed to the mild severity of these comorbidities in those patients or to the well control of these comorbidities in the state where the study was done as the presence of the tertiary level hospitals in this state indicates the easy access to high quality health care services there.
Interestingly, patients belonged to leptospirosis risk group, pregnant patients and patients with previous leptospirosis infection insignificantly spent < 7 days more frequent than their counterparts. This further supports our previous finding (not shown data) of the association of these features with the mild form of leptospirosis which is more frequently associated with shorter LOS.
The current study indicated that presence of bleeding symptoms is significantly associated with prolonged hospitalization. Moreover, logistic regression model shows prolonged PT as an independent risk factor for the prolong hospitalization (OR: 1.9). These findings underscore the importance of the close monitoring of the bleeding symptoms during the course of the disease which may indicate the need for longer hospital stay. On the other hand, conjunctival suffusion although was considered as pathognomonic symptom for leptospirosis, did not show significant association with the prolonged hospitalization and this is not surprising as the majority of cases with conjunctival suffusion spent less than seven days in the current study.
The current study indicated that hepatic involvement in form of jaundice, elevation of AST or ALT or both, and hypoalbuminemia was significantly associated with extending hospital stay. Moreover, jaundice was the strongest predictor for the prolonged hospitalization (OR: 2.67) among all the factors examined in the multivariate analysis. This finding could be attributed to the well-known severity of the icteric form which was the cause for the complications and poor prognosis in many studies [10, 41]. Also, elevated ALT was considered as an independent risk factor for prolonged hospitalization in the current study ((OR: 2) and this is not uncommon as ALT although more liver specific but its elevation indicates both hepatic and renal dysfunctions and both were seen as independent risk factors for prolonged hospitalization in the current study. On the other hand, the severe anicteric form when presents as SPHS has rapid and high rate of mortality even without presence of jaundice [42]. Moreover, rhabdomyolysis showed significant association with prolonged hospitalization. This is not uncommon as rhabdomyolysis is a main contributory factor for AKI development [43] which is considered as main cause for the severe form of the disease and its consequent longer hospital stay.
kidney was the most affected organ in the current study as its injury in form of AKI was the most common among those with prolonged hospitalization; whom more than two thirds of them suffered and significantly associated with this potentially fatal complication. Moreover, AKI was the second strongest predictor (OR: 2.28) for prolonged hospitalization after jaundice similar to another study showed increase in hospital stay was significantly associated with AKI [44]. The current study also reveals hyponatremia and hyperkalemia were significantly associated with the prolonged hospitalization and this is not unexpected as these two findings indicate further deterioration in the renal function.
Interestingly, AKI and prolonged PT which were identified as independent risk factors for prolonged hospitalization in the current study were previously found as independent risk factors for prolonged hospitalization in dengue patients as well [45, 46], which may indicate the common causes for the burden of prolonged hospitalization of the tropical diseases.
The current study showed that although late admission was not an independent risk factor for the prolonged hospitalization, but those patients with prolonged hospitalization spent significantly longer period before being admitted to hospital. Moreover, the initiation of the antibiotic after the 3rd day of symptoms onset was strongly associated with prolonged hospitalization. These two findings support our previous findings which revealed the significant association of these two delays with the development of the severe form of the disease (not shown data) which is significantly associated with prolonged hospitalization in the current study. Also, Yang, Yen [47] found delaying of starting the antibiotic therapy was correlated with prolongation of the LOS. Similarly, many studies identified the early admission and starting the proper treatment as the core components for avoiding the poor prognosis and further complications [48-50] which are expected to prolong the hospitalization. It was not unexpected to find those who were admitted to ICU had longer hospital stay (when they kept alive). This is attributed to the more and longer hospital care needed for them and the convalescent period they need after being discharged from the ICU.
The burden of the SPHS is mainly presented as higher mortality than as long hospital stay. This was not illustrated in our study and no significant association of most of the pulmonary complications with the prolonged hospitalization were observed as the predominant form in the current study was the hepato-renal form but not the SPHS.
Bajpai and Bichile [51] found that 22% of the death cases among the acute febrile illnesses in a tertiary care centre after monsoon were because of leptospirosis. Moreover, a systemic review regarding the burden of leptospirosis and its mortality estimated around 59 thousands deaths annually [52], and the large proportion was estimated to occur in young adult male in the tropical regions like south and southeastern Asia and that approaches or exceeds the number of deaths caused by other hemorrhagic fever diseases. Furthermore, mortality rates in hospitalized patients with leptospirosis range from 4 to 52% [11] and case fatality of the current study is within that range (6.5%). Early identification of the factors associated with mortality will help clinicians in advance to keep high level of suspicion to triage those patients at higher risk and immediately commencing the proper course of management needed. Moreover, considering the peculiarities of host, strain, and inoculum in different geographical areas necessitates evaluation of the risk factors of mortality in different epidemiological settings [53, 54]. No relation between the serogroup and the mortality was established, but serogroup Icterohaemorrhagiae caused much more frequent icterus and renal insufficiency than in other serogroups [55, 56]. APACHE II scores for nonsurvivors were significantly higher than survivors [57], however, severity scores were not successful enough to predict mortality in severe leptospirosis [58, 59] and the modified APACHE scoring system, which included multiple clinical and laboratory parameters was not considered as independent risk factor for mortality [60] but can help physician to start their pharmaceutical care and medical care plans earlier.
The current study is in line with the previous studies which showed old patients were significantly at higher risk for increased mortality due to leptospirosis compared to younger patients. Moreover, in one study, mortality was higher in adults than pediatrics and old age was independent risk factor for death within adults [39]. In another study, old age patients were more in need for dialysis (LOPES et al., 2004, Weeratunga et al., 2015) which was associated with the mortality in those patients. Moreover, Lopes, Costa [61] showed the mean age of their cohort was significantly higher in patients with hyperkalemia which was significantly associated with mortality. Moreover, Cetin, Harmankaya [62] found fatal cases were significantly older with significantly higher bilirubin level compared to cured patients. On the other hand, male gender although consisted the majority of our cohort (as a common feature in leptospirosis litrature) and of the fatal cases, but was not associated with the fatal cases and, surprisingly, the female gender which consisted the minority in the whole cohort and fatal cases was insignificantly more profound in the fatal cases compared to survivors. This was similar to another study showed more fatal cases among females [63]. Moreover, Everard, Edwards [64] found that death from leptospirosis was nearly twice as common among the women as among the men in Barbados. Similarly, Daher, Júnior [65] in their retrospective analysis of two leptospirosis series from 1985-1996, and from 1997-2010 observed the female gender was significantly associated with nonsurvival cases in those two different periods. However, LOPES, COSTA [39] found a higher, but non-statistically significant mortality risk in males, and Chawla, Trivedi [66] observed higher mortality amongst males compared to females in their series which had the highest mortality (52%) and the main cause of death was the pulmonary hemorrhage and the ARDS. Similarly, Antony, Celine [67] found that case fatality rate was insignificantly higher in males and was significantly increased with age in both sexes. These incompatible findings from the different studies including the current study regarding the case fatality rate in females compared to males further underscores the need for more investigation of the effect of leptospirosis on the mortality in females [68].
With the exception of pre-existent CKD, no significant association of any of the comorbidities including DM and HTN with the mortality was shown in the current study. Moreover, and unexpectedly, DM and HTN were more frequent among survivor patients. These findings are similar to other findings by Dupont, Dupont-Perdrizet [57] who found none of the studied comorbidities which included DM, HTN were significantly associated with mortality. Similarly, Sharp, García [69] found no significant association between any of the comorbidities with fatal cases including diabetes, but the HTN was insignificantly more common in the fatal cases. Findings of the current study differed partially from that study as our patients with pre-existent CKD were significantly associated with mortality (p < 0.001). Also, Sharp, García [69] showed that fatal cases did not differ significantly by reported co-morbidities or chronic medical conditions. Our finding further illustrates the unique critical impact for the previous renal insufficiency in increasing the mortality in leptospirosis patients and further allocate this group of patients as the highest risk group among all other comorbidities. This supports the ADQI 16 Workshop concepts regarding the critical effect of pre-existent CKD in developing AKI which itself associated with mortality in leptospirosis patients [70, 71].
Interestingly, our study showed obesity was not likely to increase fatality in leptospirosis patients. Although small number of obese patients was reported in the current study but none of them died. And this finding is supported by similar finding of another study which found obesity was insignificantly associated with the survivors in leptospirosis patients [69]
Similar to Abidi, Dendane [72], pneumonia is likely to increase the risk of mortality in leptospirosis patients according to data from the current report. Another study showed pneumonia was insignificantly more profound in the fatal cases. [57], but nosocomial ventilator-associated pneumonia was significantly associated with mortality [73]. Moreover, as expected, presence of sepsis was found to increase the risk of the mortality in the current report similar to another study [74], this is not uncommon as sepsis is known to cause MODs which is known to increase mortality [70, 75] and this finding is in line with Chawla, Trivedi [66] who found all the 60 patients entered the ICU had evidence of severe sepsis and more than three quarters of them developed MODs and more than half of the total patients included in that study died.
Current study showed significant association of tachypnea at admission with mortality. Also, RR > 28 was significantly associated with mortality similar to other study [60], but Abidi, Dendane [72] showed RR> 30 as independent risk factors for ICU mortality in leptospirosis patients. Moreover, Hypotension in leptospirosis can result from hypovolemia and decreased systemic vascular resistance induced by vasodilating cytokines and mediators and the current study showed significantly lower mean values of SBP and DBP at admission in the fatal cases, and around one quarter of the fatal cases were insignificantly associated with their hypotension at admission. This finding is similar to another study where the hypotension at admission was insignificantly associated with mortality [37], However, other studies showed hypotension at admission was significantly associated with mortality [76, 77].
The current study unexpectedly showed late admission was not likely to increase the risk of mortality, however, this was also identified by other studies [56, 57, 61, 78]. Moreover, interestingly, survivors were found to be admitted insignificantly later than those who were expired during their hospitalization [56, 69], however, more unexpected results were shown by Spichler, Vilaça [54] who found survivors were significantly admitted later comparing to the fatal cases. This apparently bizarre result in that study was attributed to the predominance of the severe pulmonary symptoms in the sample studied and to the fast and high case fatality in the patients with that severe form. Moreover, the current study also did not show negative consequence of the delay in starting the antibiotic on increasing mortality due to leptospirosis, and this is similar to the findings from other studies in literature which did not find significant association of the late use of the antibiotic and increasing the mortality [39, 57, 63]. However, this is in contrast to another report showed late antibiotic use was independent risk factor for mortality [79]. These unexpected findings in the current study of the absence of the significant association of delaying the admission or delaying the antibiotic initiation with increasing the mortality should be cautiously interpreted as we had high percent of the whole cohort who were late in admission (around half of all patients) and the high majority were late in starting the antibiotics leading to dissolving the statistical difference between the survivors and fatal cases. It is worthwhile to mention that benefits of early use of antibiotic in leptospirosis patients are well documented [65, 80], moreover, findings of the positive effect of using the antibiotic even in the late stage are also observed in different studies as well [61, 81, 82].
The current study also showed the longer LOS is not likely to increase the fatality cases, this is similar to another study [69], but Wagenaar, Goris [83] showed longer hospitalization period for survivors compared to fatal cases. Moreover, patients with pulmonary involvement were found to insignificantly spend longer LOS than those who did not suffered these symptoms and there was no mortality reported in the later group (Martínez García et al., 2000), however, other study showed significant longer stay of the survivor patients [37] with 18.5% case fatality rate, and in another study [76] with 13% case fatality. The shorter LOS of the fatal cases in these studies may be attributed to the severity of the cases or to the late in the admission after the onset of the symptoms [54] which make the patients at higher risk for the worse outcomes (death). Our results which showed no association of the late admission with mortality may indicate the less severity of the disease in our patients and this is supported by the less prevalence of the severe pulmonary involvement in our sample similarly to another study showed low mortality rate [84] and the less frequent severe pulmonary involvement as well. Moreover, studies showed the shorter LOS in fatal cases also showed higher fatality rate than ours indicating the higher severity of the cases in these studies which may justify the faster and higher mortality rate in those studies.
The current report showed relatively lower and slower mortality rate but with similar trend in the mortality rate after admission as with many other studies as we have the majority of the death cases were reported in the first seven days of admission and this is in line with another study [73]. This mainly attributed to the predominance of the less severe hepato-renal but not the pulmonary severe form in our cohort which is known for its higher and faster mortality as the majority of fatal cases were reported within the 24 hours of admission when that form was predominant in the cohorts studied [49, 85]. Moreover, Costa, Lopes [86] showed approximately 44% of the deaths occurred within the first three days of hospitalization and another study showed that most of the fatal cases occurred within the first five days of admission [37]. This mortality rate trend further raises the concerns and increase the need for the fast action as all studies which investigated the mortality rate in leptospirosis patients including the current study clearly showed the majority of the fatalities occur within the first few days of admission. Patrick, Rafaelle [50] attributed the absence of mortality in their cohort to the early diagnosis using the qPCR which provided unequivocal diagnosis and early treatment. This was also concluded from the observations of other studies [48, 87] which showed higher mortality in those patients who were diagnosed depending on the clinical symptoms without confirmation by MAT. This unexpected finding further encourages the searching for the predictors of the mortality in those patients to start their treatment in advance (even if the MAT showed negative results).
The current report shows lower rate of mortality (6.5%) compared to many other studies . Although we hope to avoid any death but this percent considered low comparing to other studies which showed higher mortality rate [62, 69, 79]. This relatively low rate of mortality in the current report may be attributed to the treatment measures used in these tertiary care referral centers and the familiarity of the physicians in these referral centers in endemic area (Kelantan) to early distinguish and diagnose this potentially fatal disease from other similar and less severe diseases, but this level of care and experience of practitioners may not be the same in other areas with less level of care which may lead to further delaying the diagnosis and the proper treatment which further deteriorate the prognosis and increases the mortality rate.
In contrast to many studies including the current study, Dupont, Dupont-Perdrizet [57] found hemorrhagic symptoms were not associated with the nonsurvivors. Moreover, they found the mean value of PT was significantly longer in the survivors. Presence of hemorrhagic symptoms in the current study is significantly more frequent in the nonsurvivors and this finding is similar to the finding of Durmaz Cetin, Harmankaya [88]. Moreover, Thrombocytopenia is a frequently seen complication in leptospirosis but not always associated with mortality [89], and in the current study the presence of thrombocytopenia in an equally distributed pattern between the survivors and the expired patients and the significant higher urea level and the significant association of the AKI with the expired patients suggest the uremia as the primary cause of the bleeding in addition to thrombocytopenia. Another suggested cause is the hepatic cholestasis (as our study identified a significant association of the transaminitis and insignificant association (p=0.053) of jaundice with the expired patients) which lead to prolonged PT as a result of deficiency in vitamin K dependent coagulation factors [36]. On the other hand, anemia although was more frequent with the fatal cases but was not significantly associated with increased mortality in the current study. This finding is in agreement with other studies found the higher incidence of anemia in the fatal cases [53, 62]. It is worthwhile to mention that as anemia is known to have higher prevalence in females [90] and they had higher mortality rate in the current study, it may have contributing effect on increasing mortality in females but more studies of the effect of anemia on mortality in female leptospirosis patients are needed to support this suggestion.
Goswami, Goswami [79] found all the respiratory involvement symptoms were insignificantly more frequent in the fatal cases. Similarly, in the current study most of these symptoms were more commonly observed in the fatal cases. Hemoptysis was insignificantly associated with mortality in the current study as was concluded in another study [60], and was previously reported as not associated with death although the respiratory insufficiency was considered as predictor for mortality in that study [53]. However, hemoptysis was found as significantly associated with mortality in another study [91]. Also, Esen, Sunbul [56] found respiratory symptoms were insignificantly more frequent in the expired patients, however, interestingly, they found the pathological auscultation on chest examination was insignificantly more frequent in the survivors. The current report showed the severe presentations of pulmonary involvement in form of pulmonary hemorrhage and respiratory failure although were rarely observed in current cohort similar to one study [91], but were significantly associated with the fatal cases similar to other studies [54, 72, 92]. Moreover, dyspnea and pulmonary rales were significantly associated with mortality in the current study. These findings are similar to other findings showed dyspnea was strongly associated with mortality in one study [57] and pulmonary rales were considered the same in another study [60]
Although the ECG abnormalities are commonly reported in leptospirosis patients but the clinical signs of cardiac failure are not commonly reported and although it is believed to be underestimated but will have poor prognosis whenever available [75]. In the current report, different ECG abnormalities were observed and included T-wave changes, atrial fibrillation and conduction abnormalities and all were significantly associated with the fatal cases. Moreover, cardiovascular collapse in form of shock needed inotropes which indicates the cardiac failure was also significantly associated with mortality. This is similar to other study by Pertuiset, Fen [93] who showed the cardiac arrhythmias and cardiovascular collapse as main causes of death after the pneumopathy and perfuse hemorrhage. Another study showed cardiovascular collapse was insignificantly associated with the nonservivors [57].
Jaundice in the current study was insignificantly associated with the fatal cases, and this is very common as it was not known to be fatal in leptospirosis patients in many studies [53, 57, 61]. Moreover, Spichler, Vilaça [54] found jaundice was not significantly associated with fatal cases but high level of bilirubin > 6 mg/dL was significantly associated with fatal cases in the univariate analysis but did not remain as an independent risk factor for lethal outcome in the multivariate analysis. Contrary to all previous studies including the current report, Goswami, Goswami [79] found that jaundice was significantly associated with the fatal cases. Moreover, the current study identified the elevation of any of the transaminases or both of them (transaminitis) or AST/ALT ratio may increase the risk of mortality. These findings are in line with previous studies [56, 79, 93]. Furthermore, hypoalbuminemia was associated with mortality in more than two third of the fatal cases and similar finding was observed by another study [69].
The current report as many previous studies recognizes LAKI as significantly associated with mortality regardless of its severity, moreover, it identifies statistically significant higher potassium level at admission in the fatal cases compared to survivors and hyperkalemia was more frequently observed in the fatal cases during their admission. These findings are in agreement with other findings by Lopes, Costa [61] who found that initial serum potassium was positively and strongly correlated with the risk of in-hospital death. However, Esen, Sunbul [56] identified hyperkalemia at admission as an independent risk factor for mortality in leptospirosis patients but oliguria was insignificantly more common in the fatal cases in the same study. Contrary, Sharp, García [69] did not observe elevated potassium level associated with increasing mortality neither at admission nor during hospitalization. Moreover, lower than normal blood PH was associated with increasing mortality according to data from the current study and this finding is in line with another study showed metabolic acidosis with PH <7.2 was significantly associated with fatal cases in the univariate analysis but not in the multivariate analysis [75]. However, Tantitanawat and Tanjatham [37] found metabolic acidosis with HCO3 < 20mmol/L significantly associated with severity and mortality.
Unnikrishnan, Pisharody [63] unexpectedly found that neither renal failure nor hepatic involvement were significant predictors for mortality, however the authors attributed these bizarre results to the selection bias in the sample studied. In a similar unexpected manner, Goswami, Goswami [79] found no significant difference neither in Scr level nor in urea level between the survivors and the fatal cases. But contrary, Spichler, Vilaça [54] found that even in places with a high rate of pulmonary involvement and a high rate of pulmonary hemorrhage as the major cause of death, renal failure remains an important determinant of death.
Inclusion of CNS in leptospirosis was associated with worse prognosis in many studies [70] and altered level of consciousness was found as the strongest predictor for mortality [53, 56]. However, because of the difficulty in defining that term as evident by many terms applied to altered states of consciousness by various observers like clouding of consciousness, confessional state, delirium, lethargy, obtundation, stupor, dementia, hypersomnia, vegetative state and coma, presence of any of the symptoms which indicate the inclusion of CNS may suggest the direct contribution of the CNS to the death [94]. Moreover, Pappachan, Mathew [91] identified meningism and disorientation at admission as mortality indicators. The current report showed lethargy was significantly associated with mortality in around 65% of the fatal cases. Also, disturbed consciousness although is rarely reported in the current report but shows marginally significant association with the fatal cases (p=0.066). However, Covic, Goldsmith [95] found the altered level of consciousness was significantly associated with all the deceased cases. Moreover, altered level of consciousness was insignificantly associated with mortality in the multivariate analysis in another study [57]. These findings further underscore the importance of inclusion of the neurological assessment in leptospirosis patients to accelerate the ICU admission for those with lower GCS value [22, 72]. Other CNS symptoms were rarely noted in the current report and included convulsion, encephalitis, but were insignificantly more frequent in the fatal cases in a similar manner to other studies [69, 79]
Around one third of the patients in the current study developed MODs and was not unexpected to find presence of MODs was significantly associated with mortality. This is similar to the finding of Weeratunga, Fernando [75] who showed the presence of MODs was independently associated with the mortality. In a similar manner, another study found that mortality is increased with increased the number of the dysfunctional systems and critical care management at three organ involvement stage would improve the outcome of the patients [96].
Study Limitations
Being retrospective study, we were limited to the data mentioned in the routine medical records, including the physician assessment and diagnosis for the different cases. Moreover, the relatively small number of fatal cases did not allow establishing a reliable logistic prediction model for mortality in leptospirosis patients, but the factors which positively linked with leptospirosis mortality where investigated using univariate analysis. Also, the discharge criteria may vary among the different practitioners attending patients, which may lead to different LOS depending on physicians’ opinions and this might affect the risk factors for prolonged hospitalization identified by the current study. However, patients are always discharged when their clinical and laboratory findings are stabilized. Nonetheless, the strength of this study lies on including appropriately powered sample size of leptospirosis patients. Moreover, it is the only large-scale study in Malaysia investigated the factors associated with prolonged hospitalization and mortality in leptospirosis patients.