Clinical spectrum and intermediate outcomes of community and hospital-acquired acute kidney injury: A single centre study

INTRODUCTION

Acute kidney injury (AKI) is a major public health concern, with a steadily increasing prevalence in recent decades. Its aetiology in low and middle-income countries (LMIC) is different from that of high-income countries (HIC).^1,2 Even though 85% of the world’s population resides in LMICs, data on the epidemiology and AKI outcomes from these areas are relatively scanty.³ The absence of national registries and non-uniform diagnostic criteria often results in a significant under-reporting of AKI events. Using the Kidney Disease: Improving Global Outcomes (KDIGO) or equivalent criteria, the pooled AKI estimates in LMICs are increasing and approaching those of the HICs.^3,4 Patients from LMICs are often younger, with higher relative proportions of community-acquired AKI (CAAKI). The majority do not have diabetes, hypertension, or cardiovascular risk factors, which would make them susceptible to AKI. Major inequalities exist in access to healthcare, and late presentation to healthcare systems are common. Only minimal data exist on the prevalence and causes of hospital-acquired AKI (HAAKI) in tropical countries. Hospital-acquired AKI is less common than CAAKI and is associated with lower mortality. The long-term consequences of surviving an AKI in the tropics remain primarily unknown.

India is the most populous country globally, with a high AKI burden. The aetiologic profile also shows a wide variation across urban and rural areas.^5–7 Even though conditions like acute diarrhoeal disease and cortical necrosis have decreased over the past few decades, infectious causes and envenomation remain important causes of AKI. Most data on the AKI spectrum in India stems from multiple single-centre studies. There is considerable heterogeneity in the epidemiology of AKI from different parts of India. Even in the same geographic region, important differences exist between urban and rural areas.^5–10

We studied the aetiology, in-hospital outcomes, and follow-up kidney function of AKI patients admitted to a tertiary care centre catering to a predominantly rural population.

METHODS

The study was done in a tertiary care centre in southern India. The recruitment period was from March 2017 to December 2018. All patients with AKI for whom Nephrology consultations were sought were enrolled, subject to the exclusion criteria. Patients with pre-existing chronic kidney disease (CKD) (estimated glomerular filtration rate [eGFR] <60 ml/minute/1.73m², urine albumin excretion >30 mg/day, or urine protein >+ on urine dipstick), glomerular disease, renovascular, or interstitial disease were excluded. Patients who did not have a baseline eGFR value were excluded if the ultrasound showed a bipolar length of the kidney <9 cm, loss of cortico-medullary differentiation, and cysts in the kidney. All survivors were called for follow-up visits at 2 weeks, 3 months, and 1 year after discharge from the hospital. The demographic parameters, biochemical parameters, disease severity scores, and clinical details of AKI were recorded at initial hospitalization. The treatment decisions regarding dialysis were as per the physician’s discretion. The three follow-up visits assessed the GFR using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) creatinine-based equation. A dipstick examination for urine albumin and blood pressure was recorded on each visit.

RESULTS

A total of 529 patients with AKI were recruited; 285 (53.5%) had KDIGO stage 3 AKI, 185 (23.6%) had stage 2, and 121 (22.9%) had stage 1; 288 (54.4%) had HAAKI. The AKI stages of patients with HAAKI and CAAKI have been given in Table 1. Most subjects were from rural areas (441; 83.4%). ICU admission was required in 329 (62.2%). Comorbid conditions were present in 148 (28%). A total of 195 (37.5%) required renal replacement therapy. The predominant comorbid conditions were hyper-tension (76, 14.4%), diabetes (72, 13.6%), and coronary artery disease (37, 7.0%). Baseline creatinine was 0.84 mg/dl (95% CI 0.82–0.85; available for 342 patients).

Follow-up

Among the 302 survivors, 180 came for the first follow-up visit scheduled 2 weeks post-discharge (median duration 24 days from AKI onset (IQR 20.3, 30 days). Among the 180 patients, 17 (9.4%; 10 with CAAKI, 7 with HAAKI) died in the first year following discharge. The kidney function had improved in all of them at discharge. Eight deaths occurred in patients with advanced malignancies. The cause of death was not evident in 9 patients, as it happened outside the hospital.

We had data for 156 patients beyond 3 months (135 finished all follow-up visits, and 21 patients were lost after 3 months of follow-up). Among the 156 survivors (101 with CAAKI, 55 with HAAKI), infections (n=41, 26.3%), envenomation (n=27, 17.3%), surgical AKI (n=24, 15.4%), and obstructive uropathies (n=16, 10.3%) accounted for the majority. Recurrent AKI during the same hospitalization occured in 4 patients. After a few months of recovery from the initial AKI, 2 patients had a repeat episode of AKI.

A total of 87/156 patients (55.7%) developed adverse kidney endpoints after a median follow-up of 12 months (IQR 12, 12; Table 5). Among the 10 patients with end-stage renal disease (ESRD), 5 did not recover from the initial insult and remained dialysis-dependent from the beginning. We did kidney biopsies on 3 patients with snake envenomation, suspecting cortical necrosis. Two patients had evidence of acute tubular necrosis, and 1 of thrombotic microangiopathy. The eGFR categories at the end of follow-up among the survivors have been shown in Fig. 1. The presence of HAAKI (33/55; 60%) was associated with higher odds of CKD compared to CAAKI (38/101; [37.6%]; OR 2.48; 95% CI 1.29–4.88). Those who progressed to CKD were older, with more comorbid conditions and ICU stay. Rather than the severity of kidney failure or need for dialysis, a lower eGFR on the first follow-up predicted CKD at 1 year (Table 6). A linear model with fixed effects was created to compare the eGFR trajectories of patients with and without CKD at the end of follow-up. Age, comorbid conditions, type of AKI, and ICU stay were incorporated into the model. Age (p<0.001) and HAAKI (p=0.014) were significant predictors, whereas ICU stay and comorbid conditions did not influence the eGFR trajectories. The adjusted model showed that those who developed CKD had lower eGFR from the first follow-up (Fig. 2); the recovery from the initial AKI episode was incomplete, with subsequent progression to CKD.

TABLE 5. Adverse endpoints at the end of the follow-up* (n=156)

Category	n(%)
Chronic kidney disease	70 (44.8)
New onset albuminuria	4 (2.5)
	(1 with eGFR >60 ml/minute/1.73 m2)
New onset hypertension	20 (12.8)
	(16 with eGFR >60 ml/minute/1.73 m2)

*87 patients had one or more adverse endpoints. 7 patients had more than one adverse endpoint eGFR estimated glomerular filtration rate

Close

FIG 1.

Estimated glomerular filtration rate (eGFR; ml/minute/1.73 m²) categories among survivors at the end of 1 year

Export to PPT

Close

FIG 2.

Age-adjusted estimated glomerular filtration rate (eGFR) trajectories over one year for those with and without chronic kidney disease (CKD)

Export to PPT

TABLE 6. Baseline characteristics of individuals with and without chronic kidney disease on follow-up (n=156)

Parameter	No CKD (n=86)	CKD (n=70)	p value
Age	44.3 (32.7, 52.6)	55.3	<0.001
Gender	51 (59.3)	37 (52.9)	0.516
Comorbid conditions*	10 (11.8)	32 (35.5)
Systemic hypertension	09 (10.4)	16 (22.8)	0.023
Diabetes mellitus	03 (03.4)	19 (27.1)	<0.001
Coronary disease	03 (03.4)	05 (07.0)	0.470
Stroke	01 (01.2)	01 (01.4)	1.00
ICU	28 (32.6)	30 (42.9)	0.002
Vasopressors	09 (10.5)	15 (21.4)	0.590
Mechanical ventilation	15 (17.6)	13 (19.4)	0.782
Acute kidney injury stage 3	49 (57)	46 (65.7)	0.266
Renal replacement therapy	28 (32.9)	22 (31.4)	0.841
eGFR <60 ml/minute/1.73 m2 on first follow-up	22 (25.5)	53 (75.7)	<0.001
	Median (IQR)	Median (IQR)
	– – – – – – – – – – – – –	– – – – – – – – – – – –
Hospital stay	10 (6, 17)	10.5 (7, 16)	0.807
Baseline creatinine	0.8 (0.7, 0.95)	0.78 (0.85, 0.92)	0.372
Peak creatinine	3.6 (1.8, 6.6)	2.9 (2.1, 8.4)	0.446
Creatinine on first follow-up	1.02 (0.80, 1.28)	1.77 (1.33, 2.7)	<0.001
eGFR on first follow-up	77.6 (71.8, 83.4)	40.7 (35.2, 46.1)	<0.001

*more than one comorbid conditions in some patients, total numbers will not add up ICU intensive care unit eGFR estimated glomerular filtration rate

DISCUSSION

Our study comprised 529 patients with AKI, predominantly from rural backgrounds. In contrast to the existing data from rates varying from 3.4% to 15.7%.^5–7,10 HAAKI may often be underdiagnosed, especially in surgical settings where a marginal rise in creatinine or drop in urine output can be easily overlooked.¹² In our study, the median creatinine levels in the HAAKI group were much lower compared to CAAKI. Most previous studies from India have used a serum creatinine-based definition for AKI, which might lead to underdiagnosis of HAAKI. We believe that applying KDIGO criteria resulted in higher detection rates of AKI. As we recruited patients referred to nephrology, a referral bias between medical and surgical specialties could also be a factor. Approximately 70% of subjects of HAAKI in our study were admitted to ICU, which would have resulted in closer monitoring of urine output, kidney function, and resultant early referrals.

India, our study showed a higher prevalence of HAAKI, >50% of all AKIs. HAAKI is considered uncommon in India as well as other LMICs. Previous studies from different parts of India reported prevalence The predominant aetiologies responsible for hospitalization were problems specific to the tropics. However, a few major differences in epidemiology were observed compared to data from other parts of India. Sepsis has been reported as the dominant cause of AKI in a few centres in India.^6,8,10,13 Others reported more classical tropical causes, such as acute diarrhoeal disease, envenomation, cortical necrosis, and infections like malaria, leptospirosis, and scrub typhus, as the leading causes.^5,7,9,10 Even though sepsis/infections were the dominant cause, the spectrum was dominated by necrotizing soft tissue infections occurring in non-diabetic individuals. The rural community, whose livelihood depends on farming, the hot and humid climate, and the delay in accessing healthcare facilities, may contribute to the high prevalence of skin/soft tissue infections. Soft tissue infection-related sepsis as the predominant cause for AKI was reported in another study from Tamil Nadu.¹³ Envenomation/toxins were a major cause of AKI, similar to that reported by Jayakumar et al.⁹ and Vairakkani et al.,¹³ from adjacent geographic areas. Paraquat was the most common toxin responsible for AKI, with a 90% case fatality rate. Even though around 40% of patients had a history of use of nephrotoxic medication, it could not be considered AKI’s sole cause. Most received it at the primary point of referral when other risk factors coexisted. AKI’s contribution due to diarrhoea, malaria, typhus, dengue, and leptospirosis was significantly less. The AKI spectrum shows considerable variations across India, possibly due to region-specific infections, accessibility to healthcare systems, and referral policies of individual institutions.

The occurrence of AKI due to obstetric causes has steadily declined in India, but still accounts for around 1%–5% of all AKIs.^14,15 Major regional variations in incidence are also seen. Despite being a predominantly rural population, the obstetric causes were considerably less, possibly due to referral bias. Our catchment area has a well-developed three-tier system for antenatal care, with most pregnancies being booked. Milder AKIs are often managed in secondary and tertiary care centres; those requiring life or organ support are only referred to higher centres. Internal referral bias might have contributed, as intensivists manage many obstetric AKIs, and only patients seeking a nephrology referral were included. The relatively lower proportion of diarrhoeal AKIs in our study also attests to good primary care facilities available for the public. Acute surgical conditions and malignancies were responsible for a high proportion of AKI, hitherto unreported from other single-centre studies from India.^13,16 Even though HAAKI was more common, the demographics and epidemiology appeared different from the data from developed and other developing countries.^1,17 Unlike the data from developed countries, where cardiac surgery is the most common cause of AKI, in our study, the majority underwent non-cardiac surgeries, mostly due to malignancies and emergencies.

Even though around 40% of patients had a history of use of nephrotoxic medications, the same could not be considered AKI’s sole cause. Most of them received it at the primary point of referral when other risk factors for AKI also coexisted. The contribution of AKI due to diarrhoea, malaria, typhus, dengue, and leptospirosis was significantly less in our study.

The mortality of AKI reported from Indian studies ranges from 9% to 74%.^{7,10,13,18–20} The relatively higher mortality in our study might be due to a higher proportion of ICU patients. The mortality is comparable with data from other Indian centres (40%–47%) with similar proportions of patients who need life and organ support.^6,7,13 Unlike other Indian studies on AKI, ours has a higher proportion of HAAKI; this might be another potential reason for a higher mortality. There is only minimal data on HAAKI from India; the reported mortality rates are around 40%–45%, whereas the mortality reported from general medical wards is much lower.^20–22 CAAKI appeared to be associated with lower mortality compared to HAAKI, despite having more patients with higher stages of AKI. CAAKI is reported to be associated with better in-hospital survival compared to HAAKI, despite having a more severe renal injury.^23,24 There are only limited data comparing the outcomes of HAAKI and CAAKI from LMIC countries. A recent study from Brazil reported higher mortality rates for patients who acquire AKI in the ICU but not in the wards. Two recent meta-analyses confirmed a better outcome with CAAKI, despite having comparable kidney failure and dialysis needs.^25,26 The higher mortality rates in HAAKI result from severe organ failure needing intensive care and life support rather than the severity of kidney failure per se.²⁵ We also observed that HAAKI is associated with higher mortality on univariate analysis. After adjusting for other covariates, ICU admissions, mechanical ventilation, and inotropic support were the independent determinants of mortality.

The post-discharge mortality in our study was 10%. The mortality rates tend to be higher following the recovery of AKI, with the reported mortality rates varying from 18% to 28%.^27,28

In individuals who survive a critical illness complicated by AKI, the follow-up mortality rates can be as high as 60%.²⁹ The post-discharge mortality rates appear lower, possibly due to the younger age of the population, fewer comorbid conditions, different epidemiologic profiles, and considerable loss of follow-up.³⁰ To our knowledge, no data exist on post-discharge AKI mortality rates in India. In addition to increasing short-term mortality, AKI is associated with future progression to CKD, ESRD, and adverse cardiovascular outcomes. Meta-analyses have reported 23.4–25.8 CKD events/100 person-years in those who sustain AKI.^31,32 However, the studies included in those meta-analyses did not represent the AKI profile in LMICs (most studies were from Europe and North America). The study population comprised predominantly those undergoing cardiovascular surgeries or AKI in selected subjects. The long-term outcomes of AKI in the tropics are hypothesized to be better than in the developed world due to the more favourable demographic profile: younger individuals with less comorbid conditions.³ However, only minimal data are available from LMICs to justify this conclusion. Follow-up studies are often limited to AKI resulting from a single aetiology. The reported CKD prevalence among snake envenomed patients is 22%– 41%.³³ Another Sri Lankan study reported that 9% of subjects with leptospiral AKI progress to CKD.³⁴ In our study, 45% of subjects developed CKD on follow-up. Two-thirds of those who progressed to CKD had no pre-existing comorbid conditions and were at least a decade younger than the data from the developed world. The CKD prevalence in our study is somewhat higher than the reports from HICs.^31,32 The higher CKD progression rates might have resulted from the referral nature of the centre, and receiving sicker patients with severe renal failure. A recent study from the state of Tamil Nadu reported that 20% of AKI survivors progress to higher stages; however, the proportion of patients who needed dialysis was considerably lower. The higher CKD prevalence is a matter of concern because only a quarter of the follow-up cohort had underlying risk factors. This underscores the importance of long-term periodic follow-up after AKI, which seldom happens in LMICs. We also observed that progression to CKD following envenomation was somewhat lower than other aetiologies. Schiffl et al. reported a similar observation: critically ill patients with nephrotoxic acute tubular necrosis (ATN) tend to recover better than ischaemic or mixed ATN.³⁵

We observed that rather than the severity of AKI, the recovery of kidney function acted as a determinant of long-term outcomes. Those who had a GFR >60 ml/minute/1.73 m² on the first follow-up visit had lower chances of progression to CKD at 1 year. There is accumulating evidence that initial renal recovery is an important predictor of long-term kidney function.^36–38 The risk of developing CKD is the lowest when it recovers within 48 hours, whereas 58% progress to CKD once recovery is delayed beyond 30 days.³⁸ The prevalence of new-onset albuminuria was very low in our study. Recent literature highlights the role of increased urinary protein excretion after an apparent recovery of AKI. It has been reported that there is a 23% higher chance of worsening proteinuria following one year of AKI. In addition to the GFR, proteinuria after AKI is an independent contributor to disease progression.^39,40 However, in those studies, 40%–60% of AKI survivors had pre-existing proteinuria at the time of admission with AKI. We excluded patients with baseline albuminuria and believe this might account for the low prevalence of proteinuria in our study.

The strengths of our study include prospective data collection from the time of follow-up until the end of 1 year. To our knowledge, no prior studies from India have assessed the long-term outcomes of mixed medical–surgical AKIs. The limitations include considerable loss of follow-up, especially for those with milder AKIs. There are limitations in extrapolating the data to the entire cohort, as those with milder AKI stages did not come for follow-up. Most patients were from rural areas with low connectivity; hence, logistical issues might stand in the way of scheduled follow-up visits. The study entry criteria and the intra-hospital referral policies might have influenced the aetiologic spectrum in our study. The inclusion criteria were set as patients with AKIs seeking nephrology referral. So, milder AKI cases managed by intensivists and physicians might have been excluded; this might account for the under representation of obstetric AKIs and pyelonephritis. Despite excluding patients with renal parenchymal and renovascular disorders, 44% showed evidence of CKD at 1 year. However, it might not be prudent to extrapolate these findings at community levels, as only 50% of the cohort turned up for follow-up.

We do not have biopsy data on all patients who progressed to CKD. As per the inclusion criteria, we excluded those with glomerular and interstitial pathology, so biopsies were not mandated at the time of diagnosis of AKI. We biopsied 3 patients with snake envenomation who did not recover, suspecting cortical necrosis. One patient had thrombotic microangiopathy, and the other had ATN. Those who eventually developed CKD showed partial improvement in kidney function at discharge; biopsies were not performed as the clinical setting was ATN. However, the renal parameters did not touch baseline on 1- and 3-month visits. We discussed the feasibility of biopsy in patients who did not show complete recovery, but patients did not consent, as it might not contribute to a change in treatment decisions. We do not have data on multidrug-resistant organisms contributing to sepsis. Baseline creatinine measurements were not available for all patients. Because of the poor socioeconomic conditions, people are not sensitized and seldom undergo periodic health check-ups. We attempted to exclude underlying CKD by ultrasound imaging for those without baseline kidney function. However, despite the best efforts, those with early CKD may still have been included. As subjects seeking a nephrology referral were only included, the recruitment is biased towards sicker patients with multiorgan involvement. We believe this might partially be responsible for the relatively high mortality and CKD rates. The study is not sufficiently powered to examine whether the aetiology of AKI has any bearing on long-term outcomes.