Epidemiology and costs of severe acute respiratory infection and influenza hospitalizations in adults with diabetes in India

Introduction: The incidence of diabetes mellitus is increasing rapidly in India. In addition to well-known complications, diabetes increases the risk for hospitalization and death from severe acute respiratory infection (SARI) and influenza. Here we examined the impact of SARI and influenza in Indian adults with diabetes. Methodology: This was a single-center, active surveillance study conducted in Jammu and Kashmir State, India, during the 2015–2016 and 2016–2017 influenza seasons. Adults hospitalized for SARI and receiving at least one diabetes medication were included. Demographics, health care use, and direct costs were collected from medical records and interviews of patients or caregivers. Indirect costs were estimated based on lost earnings and WHO-CHOICE estimates for hospital costs. Results: The study included 192 patients with type 2 diabetes. Median age was 66 years, median body mass index was 26.6 kg/m, and most patients had comorbidities, especially hypertension and cardiovascular disease (83.9%). Only 32.2% regularly monitored blood glucose or hemoglobin A1C, and median values at admission indicated poor glycemic control for most. Influenza was detected in 8.9% of cases. The median hospital stay for SARI was 8 days, and 22 patients (11.4%) died. Median total costs associated with hospitalization were US$710 (interquartile range, $539–$1067) for SARI patients and US$716 ($556–$1078) for influenza patients, mostly (~75%) from indirect costs. Conclusions: Adults with diabetes in India hospitalized with SARI or influenza are generally older, in poor health, and suffer from poor glycemic control. The costs for their hospitalization and care are substantial.


Introduction
The incidence of diabetes mellitus is increasing worldwide and especially in India [1].The 2017 International Diabetes Federation reported an overall prevalence of 8.8% in India, and the number of adults with diabetes rose from 32.7 million in 2000 to 72.9 million in 2017 and is projected to increase to almost 150 million in 2045 [2].This appears to be due to rapid socioeconomic development and demographic changes, coupled with increased susceptibility of the Indian population to diabetes [1].Furthermore, compared to Population form Western countries, Asian Indian people have an earlier onset of type 2 diabetes mellitus, tend to have diabetes at lower body mass indexes, convert from pre-diabetes to diabetes more rapidly, and are more resistant to maintaining glycemic control with only lifestyle changes [1,3].
In addition to well-known complications, such as cardiovascular disease, renal disease, and peripheral neuropathy, diabetes increases the risk for hospitalization and death due to severe acute respiratory infection (SARI) [4][5][6][7][8][9].This may be the result of decreased lung function caused by microangiopathy of pulmonary capillaries, autonomic neuropathy, myopathy of respiratory muscles, or changes in collagen in lung tissues [10].These risks are further increased by cardiovascular, renal, and other comorbidities common in individuals with diabetes [11].
A substantial proportion of SARIs are caused by influenza viruses [12], which circulate across India in varied seasonal patterns governed by latitude and environmental factors [13].The World Health Organization recommends that individuals with diabetes are vaccinated against influenza ahead of each season to prevent SARI and other complications [14].
To better understand the impact of SARI and influenza among individuals with diabetes in India, we analyzed active surveillance data collected over two influenza seasons (October to April) at Sher-i-Kashmir Institute of Medical Sciences (SKIMS), a 750-bed tertiary care institute in the State of Jammu and Kashmir [13,15].Here, we describe the demographics, symptoms, healthcare use, and associated costs for these patients.

Study design
This was a prospective, cross-sectional, hospitalbased epidemiological study conducted at SKIMS.The study included adult patients (≥ 18 years of age) presenting with SARI during the 2015-2016 and 2016-2017 influenza seasons (October to April) at the Accident and Emergency Medicine, Internal Medicine, Clinical Endocrinology, and Pulmonary Medicine wards.To be included, patients had to be receiving at least one oral or injectable medication for diabetes.SARI was defined according to the World Health Organization as an acute respiratory infection with a fever or history of fever (≥ 38°C) and cough within the previous 7 days requiring hospitalization [16].The primary objective was to describe the demographics and symptoms associated with SARI and influenza for adults with diabetes.The secondary objective was to estimate the costs associated with SARI and influenza hospitalizations in this population.
While hospitalized, patients were treated following the standard of care for SARI at SKIMS.Combined nasal and throat swabs were obtained from individuals with upper or lower respiratory tract symptoms and placed in 3 mL of HiViral Transport Medium (Hi Media, Mumbai, India).Samples were sent on ice or at 4°C to the laboratory within 2 hours.RNA was extracted using a QIAamp viral RNA Mini Kit (Qiagen, Hilden, Germany).Reverse transcription was performed using an AgPath-ID One-Step RT-PCR kit (Invitrogen, Carlsbad, CA, USA).Primers and probes (Invitrogen) for influenza viruses A and B were selected as recommended by the US Centers for Disease Control and Prevention and were based on genomic regions highly conserved in various subtypes and genotypes of influenza virus A (matrix protein gene) and influenza virus B (hemagglutinin gene segment).Amplification and detection were performed using a StepOnePlus Real-Time PCR System (Applied Biosystems, Foster City, CA, USA).

Data collection
Medical During a follow-up phone call 7 to 10 days after discharge, patients or caregivers were interviewed for hospitalization data (length of hospitalization; ICU admission (yes/no); cost of hospitalization to patient and insurance company; medications taken home after hospitalizations; and cost of laboratory work, imaging, and treatments) and post-hospitalization data (medications purchased after hospitalization and cost; other costs of admission; the number of working days lost by employed patients and caregivers due to hospitalizations and costs; how medical expenses were paid; cost of interest for any loans to pay for medical expenses; and receipt of influenza vaccination for the current season and costs).

Statistical analysis
Direct cost was calculated as the sum of medical costs and direct non-medical costs.Medical costs included hospital admission (including room rent, file charges, nursing charges, and attendant pass charges) medication, laboratory work, imaging, insulin therapy, and miscellaneous medical costs.Direct non-medical costs included transportation and lodging.Indirect cost was calculated as the sum of indirect hospital costs and the monetary value of lost earnings of adult patients and caregivers of all age groups due to inability to perform regular duties because of illness.Estimation of indirect cost was based on the assumption that labor was replaced at a cost to maintain societal productivity.Cost of missed work by the patient and caregiver was estimated as the product of missed work days and the median per capita income (US$ 1,861.5 per year for 2016 [17]).Missed work days were estimated by adding two days to the reported length of the hospital stay, as described by Molinari et al. [18].Indirect hospital costs were calculated using WHO-CHOICE estimates [19] as a proxy for the cost to the government in public tertiary care facilities.WHO-CHOICE estimates are available for hospital bed-cost per day in public inpatient facilities.The 2008-based year estimates were adjusted to 2016 using the consumer price index (1.93)for India [20].For inpatients, the cost to the government was the sum of the WHO-CHOICE estimates for the first day of hospitalization and the product of the WHO-CHOICE estimates for following days and (the median length in days of hospitalization − 1).Because data were not normally distributed, costs for subgroups are reported as medians and interquartile ranges.The exchange rate used was the average for 2016 (1 Indian Rupee = 0.014 US$) [21].Statistical analysis was performed using SPSS version 23 (IBM Analytics, Armonk, NY, USA).

Ethics
The study was approved by the ethics committee of SKIMS.Informed consent was obtained from all participants or their caregivers at enrolment.

Patients
The study included 192 adults with diabetes mellitus hospitalized with SARI during the 2015-2016 and 2016-2017 influenza seasons (October to April) (Figure 1).Just over half of the patients were female (56.2%), and ages ranged from 35 to 93 years (median, 66 years) (Table 1).Most patients were not employed, and just over three-quarters were non-smokers.For those who smoked, they had been smoking for a median of 35 years and smoked a median of 365 packs of cigarettes per year.
All of the patients had type 2 diabetes mellitus, and the median time since diagnosis was 4 years.Mean blood glucose was 228 mg/dL (range, 34-515 mg/dL), and the mean hemoglobin A1C level was 8.4% (range, 5.1%-12.5%).Most (94.1%) reported complying with their anti-diabetes medications, mostly insulin, although only 32.2% reported regularly monitoring blood glucose or hemoglobin A1C level.
Influenza was detected in 17 (8.9%)cases (Table 2).In most of these cases (16 of 17), the patient had not been vaccinated for influenza.

Healthcare use
Patients were admitted a median of 1 day after the onset of symptoms (Table 3).Following diagnosis of SARI, six patients (3.1%) were admitted to an ICU and the rest (96.9%) to a general ward.The median length of hospitalization was 8 days for SARI patients and 9 days for patients with laboratory-confirmed influenza.
Almost half of the patients (48.4%) had a prior visit to a hospital for the current respiratory infection, and 3.1% had a prior visit to a clinic.Approximately onethird (34.4%) had visited a non-medical practitioner (e.g.pharmacist or non-traditional healer) for the current respiratory infection.Of the 66 patients who had visited a non-medical practitioner, 64 had purchased medications following the recommendation of the practitioner.
All patients received the standard of care at SKIMS, including antibiotics, an antiviral (oseltamivir), and supportive care.Approximately three-quarters of patients (73.4%) were administered insulin, and 18.8% received hypoglycemic agents.Antibiotics were administered for a median of 7 days.Other medications include insulin, nebulized salbutamol, budesonide, heparin, diuretics, proton pump inhibitors, thiamine, and steroids.Invasive ventilation in an ICU was administered in six cases.

Outcome
Twenty-two patients (11.4%) died during hospitalization (Table 3), of which one was positive for influenza.All others recovered.All patients were prescribed medication after they went home from the hospital.

Working days lost by patients and caregivers
Employed patients lost a median of 25 days of work due to hospitalization for SARI, and a median of 30 days of work if they had laboratory-confirmed influenza.Employed caregivers lost a median of 9 days of work from caring for SARI or influenza-positive patients.

Discussion
This study, performed over two influenza seasons at a single site in Srinagar, India, showed that medical needs and costs are substantial for diabetes patients hospitalized with SARI.Medication was used in all cases, but accounted for relatively little of the overall cost.Most of the cost was instead due to indirect costs incurred by the hospital and missed work days by patients and their caregivers.The study also showed that many of the patients visited a non-medical practitioner or had a previous hospital visit, indicating that their respiratory illness had not been adequately controlled.Hospitalization due to SARI in diabetic patients amounted to a median total cost of US $710 (IQR, $539-$1067).This represents more than one-third of the median annual per capita income in India (US$ 1,861.5 per year for 2016 [17]), and suggests a substantial economic impact for these patients and their families.By comparison, for non-diabetic patients admitted for acute respiratory infections in the same Indian region, the median total cost for tertiary care hospitalization is considerably lower (US $145 [IQR: $84-279] in public settings, and US $414 [IQR: $254-771] in private settings) [22].
All patients in this study had type 2 diabetes, and most were diagnosed with community-acquired pneumonia.They were mostly old, overweight, and in poor medical condition with frequent hypertension or cardiovascular disease, all of which commonly associated with diabetes [4] and are risk factors for community-acquired pneumonia and SARI [4][5][6][7][8][9]11,23].Just over 11% of the patients died during hospitalization.This fatality rate is higher than that found in other studies of hospitalized SARI patients without diabetes (~3−4%) [24][25][26], although care should be taken with such comparisons because nondiabetic SARI admissions may be younger than our study population, and may suffer from fewer comorbidities.
Although most patients reported complying with their anti-diabetes medications, based on their blood glucose and hemoglobin A1C levels, many had poor glycemic control according to American Diabetes Association guidelines [27].This is not surprising because relatively few patients regularly monitored their blood glucose or hemoglobin A1C.
Influenza was detected in 9% of SARI cases, which is consistent with the 4−12% rates reported by other studies conducted in India and elsewhere in Asia [28][29][30].Since 2015, the Ministry of Family and Health Welfare in India has recommended influenza vaccination for individuals with diabetes [31].In support of this and recommendations in other countries, a recent systematic review concluded that influenza vaccination reduces the risk of hospitalization and mortality in individuals with diabetes, particularly those aged ≥ 65 years [32].In the current study, however, only 9% of participants were vaccinated for influenza.Similarly, in a previous study conducted between 2010 and 2012, we found that only about 9% of adults with diabetes had received influenza vaccines during the previous 1 to 5 years [33].
This study had some limitations.Importantly, the results were from a single public hospital in India, so care should be taken when generalizing them to private hospitals, where costs are expected to be higher [22], or to other regions of India or other countries.Also, our study lacked a comparator group of individuals admitted for SARI without diabetes.However, nondiabetic SARI admissions may have different characteristics (e.g., fewer co-morbidities, younger in age) that could lead to bias and limit comparisons.Another limitation is that the WHO-CHOICE estimates may underestimate indirect hospital costs [34].A further limitation is that some of the results relied on patient or caregiver recollection, so they may not have accurately remembered or accounted for costs.To help avoid this, whenever possible, we verified each expense claim using prescriptions and hospital records.However, inaccurate recollection would have had little impact on total costs.Finally, the study did not allow incidence rates to be calculated or comparisons to be made, and although data on influenza infection were collected with the goal of comparing influenza-positive and -negative cases, too few patients were positive for influenza to make meaningful comparisons.

Conclusion
This study showed that medical needs and costs are substantial for diabetes patients in India hospitalized with SARI or influenza.These findings support recommendations that individuals in India with diabetes, especially those with additional risk factors, be better monitored for diabetes control and that they and healthcare workers be made aware of the risk of respiratory infections and the need for vaccinations to prevent them.

Figure 1 .
Figure 1.Flow of patients included in the study.

Table 1 .
Characteristics of patients included in the study.

Table 2 .
Influenza and vaccination for influenza.

Table 1 .
SARI characteristics at presentation.