Studies indicate temperature related deaths are higher for cardiovascular and respiratory deaths . Our study showed an increase in cardiovascular (0.6% per°C) and respiratory (average of 2.5% per°C) mortality in Kerman with decreasing temperatures.
Many other studies have reported a delayed increase in mortality after temperature drops (cold days), lagging for a week or even more. Data from cardiovascular disease deaths from London has shown that the maximum relative risk increase of deaths for each degree C below the threshold (22.3°C) was 0.9 (95% CI 0.7 to 1.2) and occurred with a 2 to 6 day lags . In Anderson et al’s study in the US heat related mortality was associated with shorter lags (same day and previous day), whereas cold related mortality was most associated with a longer lag (current and up to 25 days previous). The reason might be the fact that cold temperatures affect mortality more indirectly than heat, such as by causing infectious disease outbreaks . However, Zanobetti et al. reported that lag 0 apparent temperature had the best model fit compared with the moving averages of multiple days .
In our study the minimum rate of mortality for both respiratory and cardiovascular deaths was during August and September. Similarly in Farajzadeh et al’s study in Tehran, the lowest number of deaths occurred in August and September. In the Tehran study the highest number of deaths had occurred in the cold months of the year (December, January, and February). Increase in cardiovascular diseases, cerebrovascular accidents, and respiratory diseases mortality all occurred during the cold months . Similar to the Tehran study, in Kerman the maximum number of respiratory deaths occurred in December, January and February; for cardiovascular diseases the maximum mortality occurred during December, January and March (Table 1).
In many studies threshold effects for temperature and mortality were seen and the relation between temperature and mortality has been reported as a J or U shape graph [7, 12]. The minimum mortality was observed at 16°C in the Netherlands, 27.5°C in Miami and 29°C in Taiwan. In another Iranian city, the capital Tehran with a more moderate climate, also a V-shape relationship was found between daily death numbers and daily temperature. The threshold for minimal mortality in Tehran was calculated as 28.5°C .
In a study in London the threshold (minimum) of total mortality and temperature was estimated to occur at about 19°C . Other studies found 20°C in New Delhi, Sao Paolo, and London and 20.5°C in Christchurch .
Temperature threshold estimates that cold-related mortality start to increase in the range from 15.8°C to 29.8°C in different studies; and for heat-related deaths have ranged from 16.8°C to 31.8°C. Interestingly, heat thresholds were generally higher in cities with warmer climates, but cold thresholds were unrelated to climate .
According to the scatterplots and fitted lines in Figure 3, increase in mortality due to high temperatures was not observed in Kerman. In our study in Kerman we observed a more close to linear relationship without a threshold. Daily mortality decreased with increases in temperature and our fitted line did not have an ascending tail in warmer temperatures. The best explanation can be acclimatization of this population over the years of living in desert climate. Acclimatization can occur through physical adaptation, special housing characteristics (materials and architecture) or behavioral patterns such as staying indoors , using appropriate clothing  and avoiding working during the hot hours. Clothing is different from the western countries in Iran. Women wear long pants and long sleeves and cover most of their skin in public. Even for adult men short pants and sleeveless clothing is rarely used in public. Traditionally, there are still old houses made of mud and straw and wind towers in Kerman which cool the house interior.
In Anderson’s study in the United States, cold effects appeared to be larger in the south (warmer climate) than in the north (colder climate). In other words, heat effects were higher in colder communities and cold effects higher in warmer communities .
Similar to our results, in Anderson’s study, heat effects were generally lower in communities with higher long term temperatures. This supports the hypothesis that communities and individuals adapt to weather even during temperatures that are warm for that area. Absolute cold effects were higher in communities with higher temperature which has been also seen in previous studies  and our study. Anderson et al. found negligible or no effect for heat in many southern US communities  and similar to Kerman, some communities did not have a minimum mortality temperature . Interestingly populations will undergo more adaptation to increasing temperatures, and less to decreasing temperatures .
Also Zanobetti et al. found a smaller risk of mortality due to high temperature in the warmer southern cities (excluding one) compared with the colder cities. Again the results were explained by the fact that persons in warmer climates tend to be more adopted to high temperature and more vulnerable to cold weather .
In our study, the pollutants such as dust and SO2 showed significant but weak correlations with only respiratory mortality. In Zanobetti et al’ study air pollution had no confounding effect on mortality . But, in another study the authors found much higher PM10 effects on mortality during warmer days . Ren et al. examined the effect of PM10 in Brisbane, Australia, and found that PM10 significantly modified the effects of temperature on respiratory and cardiovascular hospital admissions, and cardiovascular mortality at different lags . Ren et al. in another study found that ozone positively modified the association between temperature and cardiovascular mortality, with stronger temperature- cardiovascular mortality associations when the ozone concentrations where higher .
Our study was a population based study. One limitation of our study was that we were not able to examine socioeconomic and related variables such as housing conditions and availability of air conditioning which might modify the association. Also we focused on cardiovascular and respiratory disease mortality that were the most related according to some previous literature and did not examine other specific causes of mortality that may also be related . Also the effects of heat and cold on mortality may vary depending on climate factors and nonclimate factors such as diseases, sex and age .
Another limitation of this study is that we did not include humidity or influenza outbreaks. Although humidity and influenza have been accounted for as confounders in some studies , but many others have not included these variables [4, 6, 9].
Meanwhile, although some studies did adjust for humidity, they did not find a significant effect of humidity on mortality [3, 11, 12, 16, 17]. For example, Shumway et al.  showed that temperature, but not relative humidity, contributed significantly to mortality .
Also confounding by influenza epidemics, which generally occurs in the cold weather was not addressed in this study and other studies due to lack of reliable data [12, 19].
Meanwhile, it has been suggested that mortality in frailer subjects is partially because of the harvesting (or mortality displacement) mechanism, in which subjects who are likely to die soon anyway, their death proceeds by only a few days or weeks by seasonal factors such as extreme winter temperatures. However few studies have addressed this issue in their analysis . In this study due to the low daily mortality rate per day we aggregated mortality data to deaths per month, and were therefore not able to detect harvesting which needs a smaller time frame.