We found a high prevalence of vitamin C deficiency in older people in India; 74% of those in the north of India and 46% in the south of India were deficient and a further 15% and 28% respectively had sub-optimal levels. In common with other studies 
we found that vitamin C deficiency was more common in men and in users of tobacco. The lower levels of vitamin C in smokers partly reflect lower intakes but also a higher rate of ascorbate turnover possibly due to higher levels of oxidative stress in smokers 
. The association with tobacco was observed in both south and north India but the PRR was higher in the south. There were differences in the pattern of tobacco use by location. Tobacco chewing was more common in the south (28%) compared to the north (2%). Smoking manufactured cigarettes was rare in both locations (3%) but smoking beedies was higher in the north (39% compared to 9% in the south). Chewing tobacco may have a more adverse effect on ascorbate levels compared to tobacco smoking because the tobacco quid is held in the mouth for a longer period of time but experimental data are not available.
Use of biomass fuels was associated with vitamin C deficiency. The smoke from combustion of biomass fuels includes small respirable particles, carbon monoxide, nitrogen formaldehyde and polyaromatic hydrocarbons. Since many of the constituents of biomass fuels are also found in tobacco smoke 
it is likely that other adverse effects of biomass fuels on vitamin C are similar to those found for tobacco.
Seasonal differences in vitamin C deficiency varied between the north and south reflecting the different climatic and agricultural patterns across the sub-continent. In the north, the highest PRRs were observed for the main monsoon period (June to September) compared to the winter. Poor nutritional status in the monsoon months and higher dietary intakes of vegetables in the winter period have been reported from studies in the north of the sub-continent 
. In contrast the monsoon is lighter and later in the south of India. Dietary vitamin C levels also varied by season especially in the north (median of 30.6 mg/day in the winter compared to 15.7 in June to September). In the south the median values for these periods were 32.1 and 34.8 respectively.
Dietary vitamin C intakes from other studies in India 
are considerably lower than observed in western populations. However comparison of intakes of dietary vitamin C across studies and populations is limited by differences in the dietary assessment method and the availability of vitamin C data in Food Composition Tables for foods consumed in specific populations including values by cooking methods. We used a single 24 hour recall and the ICMR food composition tables which provide values of vitamin C from food items common to the Indian population. A limitation of these tables is that the values are based on raw foods. Loss of vitamin C occurs with heating and therefore dietary vitamin C in our study and in other studies in India is probably overestimated by at least 25% since the most common method of food preparation in the Indian population is cooking by heat 
. Although dietary vitamin C intakes are a major determinant of plasma vitamin C levels, this is difficult to demonstrate other than in tightly controlled experimental conditions 
. In population surveys, dietary assessment methods including 24 hour recall show only moderate correlations with plasma vitamin C. A meta-analysis of studies in high income countries reported correlations between dietary vitamin C from diet recall (ranging from one day to 12 days) and plasma vitamin C of 0.46 
. The correlation was similar for one day to longer recall, in studies excluding supplement users, and was higher in women (r
0.44) than men (r
0.36). In our study the correlation coefficient between diet and plasma vitamin C was much lower (r
0.20) and did not vary by sex. The authors of the meta-analysis concluded the moderate correlations observed might be influenced by factors including bioavailability, food processing and storage and recall errors by participants. These limitations are also applicable to our study.
Since Vitamin C is degraded by factors such as light, temperature (above 4°C) and oxidation, considerable care is required in the collection and processing of samples 
. We collected blood in subdued lighting in vacutainer tubes containing the chelating agent EDTA to prevent the continued oxidation of vitamin C from metal ions and stored the tubes in a 4°C fridge for up to 2 hours before cold centrifugation and stabilization with MPA. Greater degradation of vitamin C with EDTA compared to heparin treated samples has been reported 
. In a study of 5 people with paired samples analyzed immediately after collection using the FRASC method, the mean ascorbate was around 50 µmol/L in the EDTA samples compared to around 80 µmol/L in the heparin samples 
. Karslen et al found no significant difference between heparin and EDTA as anticoagulant when baseline ascorbate levels measured by HPLC were compared, with a mean 2.8% lower level of vitamin C in EDTA samples 
. Delayed ascorbate measurement with samples left at room temperature showed greater degradation especially for EDTA samples (e.g. 10% loss at 2 hours compared to 5% heparin). Heparin samples stored at 4°C for 2 hours showed <1% degradation in MPA acidified plasma, compared to 5.6% degradation in non-acidified plasma and 10% loss for storage for 24 hours. The equivalent data for EDTA was not collected. In contrast, Ching et al in a study of 10 people reported that samples treated by heparin, centrifuged and acidified and measured by HPLC had 7% less ascorbate than EDTA samples treated the same way 
; ascorbate loss was also significantly greater in heparin samples following a 2 hour delay in separation followed immediately by centrifugation and acidification (median 18% loss for heparin compared to 7% for EDTA). Although results from these small studies are not consistent with respect to heparin compared to EDTA and show considerable intra individual variation, all studies confirm the importance of refrigeration at 4°C for as short a period as possible, followed by immediate cold centrifugation and acidification. Once samples are frozen, plasma ascorbate is stable over long term storage at −70°C 
. In our study the median storage time was just over one year. Although we had a clear protocol for the collection and processing of samples and laboratory staff were trained to follow the protocol, we cannot exclude that errors may have occurred leading to loss of vitamin C from the samples. Vitamin C showed typical patterns observed consistently in other studies, such as lower levels in men, in tobacco users, those with indices of poor nutrition, lower socio-economic status and an inverse association with age 
. It is unlikely that these patterns would be preserved if the blood samples had degraded randomly but a systematic loss would lead to an over estimation of the prevalence of vitamin C deficiency. Quantifying the possible ascorbate loss in our study is uncertain but based on the literature reviewed above 
we might expect only minor degrees of ascorbate loss (possibly up to 10%) due to pre-analytical factors such as use of EDTA, and delays in centrifugation and freezing of samples in view of the sample handling protocol described in this study. However we acknowledge that the losses might be greater since we did not have any formal methods of quality assurance to ensure the protocol was followed. The levels for participants in north India in the present study were very similar to those in a small feasibility study we conducted previously in Haryana 
We had only a single measurement of plasma and dietary vitamin C and were unable to ascertain the effects of within person seasonal changes. Our response rates were acceptable (75%) and apart from age there was no response bias in sex or socio-economic status. Since vitamin C deficiency increased with age the prevalence of vitamin C deficiency might be underestimated.
Our population was primarily rural or from small towns, characterized by low BMI, high tobacco and biomass fuel use and low intakes of dietary vitamin C. In 15% the mid-upper arm circumference values were indicative of moderate to severe malnutrition. Our results may not apply to middle aged and younger people, city dwellers or high income groups and studies are required in these groups.
In addition to low dietary intakes of vitamin C, low plasma levels of vitamin C in India may also reflect haptoglobin (Hp) allele status (Hp1 or Hp2). The Hp2 -2 phenotype is substantially higher in India (around 70–80%) compared to populations of European ancestry (30–40%), and conversely Hp1-1 is much lower, less than 3% in India compared to around 15–20 % in Europeans 
. Studies in Europeans have reported around 20% lower plasma vitamin C levels in those with the Hp2-2 polymorphism compared to those with Hp 1-1 
. A study of University of Toronto non-smoking students found vitamin C deficiency in 17% of those with Hp2-2 compared to 11% of those with either Hp 1-1or 2-1. The risk of deficiency in those with low dietary intakes of vitamin C (below recommended intakes) was modified by Hp status; from an OR of 1.7 for Hp1-1 or 2-1 to an OR of 4..8 for HP2-2 
. These data suggest that the effect of Hp2-2 may be greatest when dietary intakes of vitamin C are low. No data are presently available in India on vitamin C and haptoglobin polymorphism. An important function of haptoglobin is to bind haemoglobin preventing peroxidation by free iron; the observation of lower vitamin C levels in Hp2-2 individuals with lower haptoglobin may reflect the increased depletion of vitamin C due to reduction of free iron 
. Haptoglobin polymorphisms might also explain in part the lower levels of vitamin C reported for South Asians in the UK compared to those of European or African ancestry 
or for Indians in Singapore compared to Chinese 
. However the differences in plasma vitamin C between ethnic groups in these studies were not large and the mean levels were much higher than in our study population. The studies on vitamin C in Indian ethnic groups in Singapore and the UK are in the settings of high income countries. Indians in these settings are characterized by BMIs in the normal to overweight range, more central obesity and higher dietary energy intakes. Although data are sparse it is likely that dietary intakes of vitamin C in Indians are also higher in high income settings, probably reflecting better nutrition of the Indian ethnic groups in contrast to our study participants. In a study in the UK, children of Indian ethnicity had slightly lower dietary intakes compared to white European children but both groups had intakes well above the recommended intakes for their age group 
. Currently there are limited data on other genetic modifiers of vitamin C levels 
and no studies have been carried out in India.
The majority of previous reports on vitamin C deficiency from population based studies have taken place in the UK or North America. In these studies the prevalence of vitamin C deficiency ranged from 26% of men and 14% of women aged 25 to 74 years in the Glasgow MONICA study
,25% of men and 16% of women aged 19 years and over in the UK Low Income Diet and Nutrition Survey
, 14% of nonsmoking women and men aged 20–29 years in the University of Toronto campus 
. In two waves of the US Nutrition and Health Examination Study of people aged 20 years and over, vitamin C deficiency was reported in 18% of men and 12% of women for 1998–1994 
, reducing to 10% and 7% in the 2003–2004 survey; the prevalence for those in the lower income groups was double that of the high income groups at both time periods 
. Only two studies have been conducted outside high income countries including one from India. A nationally representative population study from Mexico reported a 40% prevalence of deficiency in women of childbearing age 
. Men and older people were not included in the study. In a small study of 322 people aged 20–50 years from western India, vitamin C deficiency was found in 9.6% of men and 13.0 % of women, and just over a half had levels in the sub-optimal range 
In conclusion, we found vitamin C deficiency in a substantial proportion of the older population in two settings in north and south India. Only 10% of those in the north and a quarter of those in the south met the criteria for adequate levels. Our results are relevant to current debates about the control of non-communicable diseases in India. Low fruit and vegetable intake, tobacco use and biomass fuels contribute respectively the third, fourth and fifth ranked risk factors associated with mortality and disease burden in India 
. Our results show that low dietary vitamin C intakes (reflecting low fruit and vegetable intake), tobacco use and biomass fuels are risk factors for vitamin C deficiency and add to the evidence on the health consequences of these risk factors. In poor communities, such as described in our study, consideration needs to be given to measures to improve the consumption of vitamin C rich foods and to discourage the use of tobacco. This includes a raft of measures including agricultural and tobacco policy, promoting awareness in communities through education and employment of local dieticians. The growing proportion of older people in India also highlights the importance of better information on the nutritional status of this age group.