Survivability of hospitalized chronic kidney disease (CKD) patients with moderate to severe estimated glomerular filtration rate (eGFR) after experiencing adverse drug reactions (ADRs) in a public healthcare center: a retrospective 3 year study

Background Accurate identification and routine preventive practices are crucial steps in lessening the incidence of medications and patients related adverse drug reactions (ADRs). Methods Three years retrospective study was conducted among chronic kidney disease (CKD) patients at multi-wards in a tertiary healthcare center. Data collected included demographic characteristics, physical examination results, comorbid conditions, laboratory tests and medications taken. Only medication prescribed during the hospital stay were considered in this study. Results From this study only one ADR incident was definitely preventable and majority of other ADRs (88.3%) were possibly preventable. Type of renal replacement therapy (p = 0.023) and stages of renal function (p = 0.002) were significantly associated with survivability of the hospitalized CKD patients after ADRs. Highest percentage of mortality based on categories were 50–59 years (20.0%), male (16.3%), Indian ethnicity (23.7%), obese (15.0%), smoking (17.1%), consumes alcohol (17.4%), conservative management of renal disease (19.5%) and renal function of < 15 mL/min/1.73m2. Overall survivability using Kaplan-Meier analysis reported a significant difference of 18-day survival rate between patients undergoing hemodialysis and patients conservatively managing their renal disease. The 18 days survival rate of patients undergoing hemodialysis, peritoneal dialysis and conservative management were 94.9%, 91.7% and 75.1% respectively. Eighteen days survival rate of patients with renal functions of 30–59 mL/min/1.73m2, 15–29 mL/min/1.73m2 and < 15 mL/min/1.73m2 were 87.4%, 69.8% and 88.6% respectively. Similarly, Cox regression analysis revealed that renal replacement therapy was the only factor significantly contributed to ADRs related mortality. CKD patients whom conservatively managed renal disease or/and with renal function of < 15 mL/min/1.73m2 had 5.61 and 5.33 higher mortality risk respectively. Conclusion Majority of the reported ADRs were possibly preventable. Renal replacement therapy and/or renal function were significant risk factors for mortality due to ADRs among hospitalized CKD patients stages 3 to 5. Clinician engagement, intensive resources and regular updates aided with online monitoring technology are needed for enhancing care and prevention of ADRs among CKD patients.

developed a scale for assessing the severity of ADRs. Classification on severity are mild (slightly bothersome; relieved with symptomatic treatment), moderate (bothersome, interferes with activities; only partially relieved with symptomatic treatment) and severe (prevents regular activities; not relieved with symptomatic treatment) [14][15][16]. ADRs preventability are determined by ADRs types which ranges from type A till type D. Type A or Type 1 (augmented) reactions results from an exaggeration of a drug's normal pharmacological actions when common therapeutic dose administered. Type A is usually dose dependent. Type B or Type 2 (bizarre) are ADRs that occurs as novel response not expected from known pharmacological action [17]. Type C (chronic) ADRs includes adaptive changes, rebound phenomena and other long-term effects. Type D (delayed) reactions are carcinogenesis, affecting reproduction such as impaired fertility and adverse effects on the foetus during early or later stages of pregnancy and drug availability in breast milk [16].
Chronic kidney disease (CKD) is a major health burden that amplifies the risk for adverse events [18,19]. CKD is independently associated with increased adverse risks including kidney failure, cardiovascular events and all-cause mortality [19,20]. An eight-year (1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006) retrospective study conducted in the United States revealed that there were more than 2 million deaths which were attributed as ADR-related deaths [3]. Additionally, Pirmohamed et al. [21] reported high incidence of in-hospital ADRs which was about 14.7%. Therefore it is beneficial to evaluate patients risk factors for ADRs individually. For minimization of ADRs events, understanding and knowledge on prescribed drug metabolization mechanisms, magnitude and probable ADRs is essentially to be equipped by the healthcare professionals. Thus, these will establish safe medication prescription practices which stresses cautious consideration of the benefits and risks of concomitant medications [22]. Therefore, this study aimed to assess the causality, severity and preventability of ADRs among hospitalized CKD patients with estimated glomerular filtration rate (eGFR) value of < 60 ml/min/1.73m 2 . Additionally, risk factors for mortality due to ADRs were also evaluated.

Study design and participants
This a 3 years retrospective observational study conducted in Penang General Hospital. It is the seconds largest General Hospital in Malaysia. A total of 1070 medical records of patients experienced ADRs from various wards for the duration of 3 years (January 1, 2014 till December 31, 2016) were screened. CKD patients stages 3-5 (eGFR< 60 ml/min/ 1.73m 2 ) with stable serum creatinine (sCr) values during the initial days of admission and experienced ADRs during hospitalization were the primary inclusion criteria of this study. The sCr value obtained during the first day of admission were used to estimate the glomerular filtration rate (GFR). Additional inclusion criteria were patient aged ≥18 years old and admitted for more than 24 h. Medical records which were dubious and incomplete and ward admission due to ADRs or acute kidney injury (AKI) were excluded from this study. Only 160 patients were selected after subsequent screening and identification of records that met the inclusion and exclusion criteria. From the total number of the patient records finally selected, 132 patients survived and 28 patients did not survive ADRs during hospitalization. Prior to study commencement, ethical approval was obtained from Medical Research & Ethics Committee (MREC), Ministry of Health Malaysia (MOH). Study approval number: NMRR-15-1810-28,375(IIR).

Estimation of renal function
For each patient, the sCr value was measured at admission using the standardized GFR method in the hospital laboratory department. The eGFR was then calculated from serum creatinine value using the chronic kidney disease epidemiology collaboration (CKD-EPI) equation [23]. Stages of CKD included in this study were 3A eGFR 45-59 mL/min/1.73 m 2 , 3B eGFR 30-44 mL/min/1.73 m 2 , 4 eGFR 15-29 mL/min/1.73 m 2 and 5 eGFR < 15 mL/min/ 1.73 m 2 . Based on the type of renal replacement therapy, patients with end stage renal disease (ESRD) were divided into hemodialysis, peritoneal dialysis and patients not undergoing any type of dialysis (will be termed as 'conservative management or conservatively managed renal disease' in subsequent sections).

Data collection
For each patient, data was collected retrospectively from the patients' medical records using a standardized form. Data collected included (a) demographic characteristics such as age and sex; (b) physical examination results such as blood pressure and weight (c) comorbid conditions such as diabetes, hypertension, vascular disease, heart failure, atrial fibrillation and anemia (d) laboratory tests such as serum and biochemical parameters and (e) medications taken before admission, during hospitalization and medications prescribed at discharge. Only medications prescribed during the hospitalization were considered in this study.

Identifications of ADRs
The primary outcome of this study was to determine the incidence and patterns of ADRs among hospitalized CKD patients stages 3 to 5. Identification of adverse drug reaction (ADR) event was done using a 3-step identification process (trigger list/ physician order, confirmation by an independent reviewer and assessment of causality, severity and preventability of identified ADRs by experienced pharmacist). In this study, ADRs was defined according to Edwards and Aronson [24]. Suspected ADRs were then classified based on the system developed by Rawlin and Thompson [25]. For each suspected ADR, information collected were (a) date start and end of ADR (b) the probable ADRs causative drugs, administered dosage and frequency (c) physical examination and laboratory results (d) reported adverse outcomes such as dizziness and rash. The beginning of the ADR was the date of the clinical or biological diagnosis of the ADR. The end of the ADR was the date of normalization of the effect which was obtained from the ADR reporting form and justified with the date of laboratory examination with normal results or the disappearance of clinical symptoms reported by physicians and pharmacist. If the end date of ADR and the date of the patient demise was reported on the same day therefore the cause of death is regarded as due to ADRs. The ADRs lasted from 1 day to several weeks. Major drug classes that attributed to ADRs were anti-infective, anti-hypertensive, analgesic, statins and anti-diabetic. Furthermore, anti-infectives contributed to highest number of mortality in this study (Danial M, Hassali AMA, Ong LM and Khan AH. Direct cost associated with adverse drug reactions among hospitalized chronic kidney patients in a public healthcare facility: A retrospective 3 year study, submitted).

Assessment of causality, preventability and severity of ADRs
Assessment of ADRs were done based on causality, severity and preventability. The drug related causality was assessed by using Naranjo algorithm [26]. Only definite and probable ADRs were considered for further assessment. The severity of ADRs were then scored using Hartwig and Siegel [13] scale into mild, moderate or severe. Preventability of ADRs were determined using Hallas et al. [27] criteria into definitely preventable, possibly preventable and non-preventable. The overall incidence of ADRs were defined as the total number of patients who suffered ADRs during hospitalization in relation to the total number of patients admitted to various wards during the 3-year study period.

Statistical analysis
For the purpose of descriptive analysis, baseline characteristics of patients with ADRs were analyzed using either chi-square test for categorical variables and t-test or Mann-Whitney test, depending on the skewness of data, for continuously distributed variables. The Cox regression analysis was used to estimate of the relative risk of having an ADR during hospital stay in relation to stages of renal function (stages 3 to 5) or for ESRD in relation to the three types of renal replacement therapy (hemodialysis, peritoneal dialysis and conservative management). The Cox regression model is the most frequently used model for analyzing time-to-event data [28]. In this case, the time from hospital admission to day on which the ADR occurred was considered as the time to event and the outcome of the model is either survival or death. The advantage of using Cox regression model is the ability to censor patients who fail to reach the study end-point [29]. In this case, patients who survived of an ADR during the hospitalization were censored. The hazard ratio is the probability that a patient survived the event or the outcome to a certain time point [29]. The hazard ratio survival of the ADR event was reported graphically using the Kaplan-Meier estimates, plotting the logminus survival function over time. The log-rank test was used to investigate the association with the outcome. All analysis was performed using SPSS (version 22; SPSS Inc., Chicago, IL). Two-sided p-values of less than 0.05 were considered statistically significant.

Baseline characteristics of patients
Baseline characteristic of patients were similar as reported in (Danial M, Hassali AMA, Ong LM and Khan AH. Direct cost associated with adverse drug reactions among hospitalized chronic kidney patients in a public healthcare facility: A retrospective 3 year study, submitted). CKD patients were grouped into survived (n = 132) and whom did not survive after ADRs (n = 28) event during hospitalization. Majority of the study patients were Chinese; male; aged ≥60 years; with eGFR value of < 15 mL/min/ 1.73m 2 and conservatively managed the renal disease (Table 1). Furthermore, the CKD patients were reported to have comorbidities primarily such as diabetes, dyslipidaemia and hypertension. Additionally, it was reported that they consumed ≥23 of total number of medications (Danial M, Hassali AMA, Ong LM and Khan AH: Development of a mortality score to assess risk of adverse drug reactions among hospitalized patients with moderate to severe chronic kidney disease, submitted).

Kaplan-Meier overall survivability
The Kaplan-Meier overall survivability analysis performed indicated 85.0% survival for the duration of 18 days (Fig. 2) (Table 5).

Cox regression
The Cox regression analysis revealed that only renal replacement therapy contributed significantly to mortality associated with ADRs among CKD patients. Factors such as age, gender, ethnicity, BMI, smoking status, alcohol consumption status and renal function were not significantly linked to mortality due to ADRs among CKD patients

Discussions
Identification of ADRs among CKD patients will be useful in clinical practice as to implement appropriate care aimed at reducing the number of ADRs. Survival estimation studies are vital for the prediction of impending disease burden, redirection in approaches of disease screening and planning of clinical trials both intervention and observational studies, thus paving ways for more successful understanding among public, healthcare providers and policy makers [30]. Based on our study, in Penang General Hospital yearly about 13.5% of hospitalized CKD patients stages 3 to 5 experienced ADRs. Our findings were similar with metaanalysis study conducted by Lazarou et al. [31] where they reported that 10.9% of patients experienced ADRs of all severities as inpatients and another study by Davies et al. [32], which estimated that between 10 and 20% of patients experienced ADRs during hospitalization. Moreover, from our study about 88% of ADRs were possibly preventable. Similar results were reported by Chan et al. [33], where it was reported that 50% of ADR were preventable. Preventable ADRs were commonly associated with prescription of diuretics, antiplatelet, anticoagulant, antidiabetic and NSAIDs drugs to the patients [34,35].
Severity assessment using the Hartwig and Siegal indicated that nearly 27% of ADRs scored mild and nearly 60% of ADRs scored at level 3 or below on the Hartwig scale. This indicated that remedial action was performed to treat the ADRs as reported in the patients' medical records. The reported remedial action was either discontinuation of the suspected drug alone and/or treating with the corrective drug. The outcome of the remedial action resulted in additional laboratory investigations, extra procedures, increment in days of hospitalization, admission to intensive care unit and/ or death.
Least number of survivors were from age group of ≥60 years (84.2%) using Kaplan-Meier analysis. It has been reported that death is anticipated after an ADR among patients aged more than 55 years [31,35]. It is attributable to the presence of high levels of albuminuria with impaired level of eGFR among the older adults [30,36].
In addition, in this study males were more susceptible to mortality due to ADRs compared to female. The Kaplan-Meier survival analysis revealed that the lowest survival experience for male CKD patients to be 84.7%. Inker et al. [30], indicated that the differences between gender may be attributable to faster progression to ESRD in male compared to female, inaccuracies in estimating equations or lower levels of normal GFR in women [37,38]. For example, Fan et al. [39] reported that the CKD-EPI equation has slight but non-significant overestimation of GFR in women compared with men. Furthermore, gender differences may also due to immunological and hormonal physiology which influences pharmacodynamics and pharmacokinetics responses, particularly in relation to cardiac and psychotropic medications [40]. In this study, significant positive association were found between renal function and survivability after ADR events (r = 0.02, p < 0.05). Patients whom conservatively managed renal disease (19.5%) and patients with renal function of < 15 mL/min/1.73m 2 were at the highest risk of mortality. Metabolic changes among advanced CKD patients have been associated with impaired physical function, which includes reduced muscle function, grip strength and cognition [41]. Furthermore, poorer degree of CKD is associated with higher frequency of the frailty syndrome and higher risk of functional decline over time [42][43][44].
Safe drug dosage is specifically influenced and associated with individual factors like physical parameters including age and weight; presence of comorbidities; physiological status including renal and hepatic function; current medications usage and previously reported history of allergies [45]. Age influences drug pharmacokinetics and pharmacodynamics activities [46]. Older age influences the rapid accumulation of total body fat and inversely reduces lean muscle mass and water volume. Thus, it impairs the dissemination of many drugs for example benzodiazepines, antipsychotics and opioids [47]. Additionally, reduced water availability rises toxicity levels. This condition results in lengthening of drug elimination half-life which in turn causes undesirable drug side effect such as drowsiness, falls and unwanted dosage build-up [47]. Likewise, aging causes impairment of GFR like decrease in renal size, nephron function and assimilations which are accountable for comorbidities such as hypertension, diabetes and heart failure [48]. Therefore, estimation of GFR is vital when up-taking renally eliminated drugs like dabigatran and metformin as safe dosage needed to be prescribed as to lessen the risk of an ADR [48]. It also have been reported that aging causes decline in liver mass and perfusion, which can adversely affect drugs with high hepatic extraction ratio such as diltiazem, opiates and warfarin [49]. These drugs systemic bioavailability surges with higher accumulation in serum coupled with enhanced adverse drug effect [47].
Pharmacovigilance of ADRs includes detection, assessment, understanding and prevention of adverse effects or any other drug-related problem with the aim of enhancing medication safety and patient care [50]. Likewise, E-pharmacovigilance will be a resourceful tool in drug safety for potentially predicting an ADR likelihood by utilizing previously obtained information such as laboratory investigations [51]. Moreover, drug safety can be established by adopting programmed electronic methods which can render information on past errors of medication and/or dosage and potential medication interactions. Beneficial attributes of electronic prescribing methods have been applauded by a recent systematic review where it was reported to reduce medical errors and adverse drug effects [47,52].

Study strengths and limitations
The strength of the current study lies in identifying the ADR event using the 3-step identification process. The study population included all ADR events recorded for 3 years continuously from multiple wards, representing all clinical specialties commonly found in most acute hospitals. Furthermore, the age distribution of our study population was comparable to figures for all in-patient admissions from other literatures. Thus, this study produces highly reliable results that represents the real-world practices. However, the study limitations were firstly, it was conducted in one hospital and there is likely to be variation between different hospitals because of differences in the local population characteristics and the specialties within the hospitals. Secondly, since this study was performed in one hospital, it may limit the generalizability of the results. Finally, survival risks for specific stages may be overestimated because it was derived using only one measurement of GFR.

Conclusions
Conclusively, from this study only one ADR incident was definitely preventable and majority of other ADRs were possibly preventable. Renal replacement therapy and/or renal function were significant risk factors for mortality due to ADRs among hospitalized CKD patients stages 3 to 5. There is a need to develop a high-reliability assessment tool which can meticulously establish suitable diagnostic criteria for ADRs with universal acceptance to improvise the fundamental aspect of drug safety and evades the impending ADRs.