Diabetic nephropathy
Sir
I read with great interest the article by Afifi and colleagues [1] in the Eastern Mediterranean health journal and I appreciate the authors’ work. However, I would like to make a few comments because of the increasing prevalence of diabetes and its complications in the Arab world. Diabetic renal disease is a common complication and is the most prevalent cause of end-stage renal disease (ESRD) in the Western world [2]. According to the ESRD programme in the United States, the number of existing patients with terminal renal insufficiency caused by diabetes more than tripled between 1990 and 2001 [3].
The first question to pose here is: was this a 6-year study? The title possibly suggests it is a cohort study. However, it seems from the study design, tables and the results that it is a data review of an ambiguously selected sample of ESRD patients enrolled in the Egyptian renal data system. Both the sampling technique and the rationale behind the sample size are unclear. Moreover, in a cohort study we trace a risk factor to get an outcome, either prospectively or retrospectively. Here, the authors already know the outcome, i.e. ESRD, and the risk factor, diabetes, and the rates shown by the data are not in line with what cohort studies are supposed to reveal.
Moreover the results seem to be contradictory. The authors say that “the mean age of patients with diabetic nephropathy was higher than that of patients with other causes of ESRD in the years [they] studied”. They then say that “mortality among diabetic patients with ERSD was higher than in patients with ERSD from other causes”. If we want to know how long patients with ESRD due to diabetic nephropathy live in comparison to ESRD due to other causes, we must evaluate the interval between 2 events: the diagnosis of ESRD, whatever its cause, and death. In fact, the 2 events of interest – diagnosis and death – may not occur for all the study participants during the period they are observed and the actual period of observation may not be the same for all of them. These complicating factors rule out the possibility of simply calculating the average time between the 2 events or the mean age of the studied participants as the authors did in Table 3, where they compared the mean age of diabetic nephropathy patients and patients with ESRD from other causes. A special statistical technique, survival analysis [4], is needed to look at the interval between diagnosis and death when the first timing is not the same for everyone, the second does not necessarily happen to everyone, and hence the study participants are observed for different periods of time.
References
- Afifi A, El Setouhy M, El Sharkawy M, Ali M, Ahmad H, El Menshawy O, Masoud W. Diabetic nephropathy as a cause of end stage renal disease in Egypt: a six year study. Eastern Mediterranean health journal, 2004, 10(4/5):620–6.
- Parmar MS. Recurrent hypoglycaemia in a diabetic patient as a result of unexpected renal failure. British medical journal, 2004, 328:883–4.
- Excerpts from the United States Renal Data System 2003 Annual Data Report: atlas of end-stage renal disease in the United States. American journal of kidney disease, 2003, 42(6 suppl. 5):S1–226.
- Norusis MJ. SPSS base system user’s guide, Release 5.0. Chicago, SPSS Inc., 1990.
M. Afifi
Department of Research & Studies
Directorate General Planning
Ministry of Health
Muscat
Oman
Correspondence to:
Authors’ response
We are glad that our work stimulated Dr Afifi from Oman to send his views. We believe that if he had read the article more thoroughly, he would have found answers to most of his criticisms. However, we would like to clarify the points he raised.
There was no need to repeat the extent to which diabetes is responsible for chronic kidney disease in the United States and Western countries. We, as well as all nephrologists are well aware of this issue and have quoted it in our work. We would like to draw his attention to the fact that what applies to Western countries does not necessarily apply to developing countries for a variety of reasons he should know. Indeed, differences exist between developed and developing countries and we should always investigate and analyse our problems rather than quoting data from the United States and Europe. In this respect he should read Table 1 that showed variations in the prevalence of this disease ranging from 4.7% in some countries to 50% in others.
Dr Afifi postulated that our study was a cohort study although we never said that it was. We clearly stated that it was a six-year multiple cross-sectional study. The writer should know that there are many limitations to a cohort study in Egypt and in most developing countries as well. The most important of these limitations is the unavailability of complete medical records and the reluctance of colleagues to share their records. We would be happy to share in data management and analysis with the writer if he has such data.
He suggested that we have contradictory results because “the mean age of patients with diabetic nephropathy was higher than that in patients having other causes of ESRD” and that “mortality among diabetic patients with ERSD was higher than in patients with ERSD from other causes”. There is nothing contradictory in these findings. Indeed, the mean age of diabetic patients with ESRD was about 10 years higher than that due to other causes (Table 3) and this is simply explained by the fact that type 2 diabetes needs many years to develop into ESRD, whereas other causes of ESRD, particularly infections causing glomerulopathies, occur at a much younger age. The mortality among diabetics was also much higher than in other causes of ESRD (Table 2). This does not need much explanation as we are sure that the writer agrees that diabetics suffer from increased mortality due to micro- and macrovascular complications.
A. Afifi
Department of Internal Medicine and Nephrology
Ain Shams University
Cairo
Egypt
Correspondence to:
M. El Setouhy
Department of Public Health
Ain Shams University
Cairo
Egypt
Risk factors for hypertension in UAE
Sir
I have read with great interest the article by Sabri and colleagues [1] in the Eastern Mediterranean health journal and I appreciate the authors’ work. However, I would like to make few comments because of the importance of hypertension and its risk or associated factors.
The authors mention in their abstract as well as in the results and discussion sections that hypertension among cases was high in low-income participants. Looking at the bivariate analysis (Table 1) and the multivariate analysis (Table 4), we would see the reverse. Table 4 shows that those with higher income (income equal 5000 dirhams or above) were 2.69 times more likely to be hypertensive than others.
The authors contradict themselves again where they show that the crude odds ratios for obese participants (BMI ≥ 30 kg/m2), those with no physical activity, and those with positive family history of diabetes were 0.46, 0.72 and 0.28 respectively (Tables 1 and 2). This suggests these factors are protective against hypertension, whereas in Table 4 the adjusted odds ratios for the previously mentioned variables are 4.29, 1.8 and 2.58 respectively, suggesting that obesity, sedentary lifestyle and family history of diabetes are risk factors for hypertension.
The cause behind such contradiction is the order of rows in the Tables 2 and 3 from which the odds ratios were calculated in the bivariate analysis. The correct order is to put the exposed or the risk factor in the first row and the non-exposed in the next [2,3]. If the rows are reordered in the bivariate analysis tables, the correct odds ratios for variables studied are obtained and these would be 1/existing figure.
The authors mention in the sampling section that cases and controls were age, sex and nationality matched. They say that they recruited 500 hypertensive patients, 64 did not participate and thus 436 cases and 436 matched controls were included in the final analysis. From Table 1 the number of cases and controls varied markedly in sex, age group and slightly in nationality. The problem here is that hypertensive patients under 40 years were more reluctant to participate in the study in comparison to those aged 50 years and above where cases outnumbered controls. Could this raise the question of selection bias? Al-Shahi et al. [4] investigated the differences between consenters and non-consenters in demographic and clinical features at first presentation and outcome during follow-up. They found differences between adults who consent to participate in observational record-based research and those who do not or cannot. Accordingly, they recommended that further research should be directed towards exploring consent bias in other disease groups and other research designs to see if the bias is pervasive and remains unpredictable.
As regards the study tools, it is slightly ambiguous whether the authors used a self-reported questionnaire or a structured interview. Did the authors use a self-reported questionnaire to be filled by literate participants and resort to a structured interview for illiterate participants?
Finally, drawing the conclusion that “more expatriates than UAE nationals had hypertension is in keeping with previous findings of a Swedish study” (page 618), is flawed [5]. First, the difference in the Emirate study was not significant even in bivariate analysis (Table 1). Moreover, the Swedish study adjusted the foreign-born individuals for sociodemographic and lifestyle factors, which was not done in the Emirate study.
References
- Sabri S, Bener A, Eapen V, Abu Zeid MSO, Al-Mazrouei AM, Singh J. Some risk factors for hypertension in the United Arab Emirates. Eastern Mediterranean health journal, 2004, 10(4/5):610–9.
- Greenberg RS et al. Medical epidemiology, 2nd ed. Stamford, Appleton & Lange, 1996:126–8.
- Fletcher RH, Fletcher SW, Wagner EH. Clinical epidemiology: the essentials, 3rd ed. Baltimore, Williams & Wilkins, 1996:215–8.
- Al-Shahi R, Vousden C, Warlow C; Scottish Intracranial Vascular Malformation Study (SIVMS) Steering Committee. Bias from requiring explicit consent from all participants in observational research: prospective, population based study. British medical journal, 2005, 331(7522):942.
- Sundquist J. Living conditions and health. A population-based study of labor migrants and Latin American refugees in Sweden and those who were repatriated A population-based study of labor migrants and Latin American refugees in Sweden and those who were repatriated. Scandinavian journal of primary health care, 1995, 13(2):128–34.
M. Afifi
Department of Research & Studies
Directorate General Planning
Ministry of Health
Muscat
Oman
Correspondence to:
Authors’ response
Dr Afifi has raised some important points concerning our study [1]. We used a matched case–control design [2–4]. The method we employed was the one that adjusted for age, sex and nationality [5]. It is true that the association of risk factors may vary by age, sex and nationality, which are mostly fixed factors, but so long as the sample is adjusted equally among the cases and controls with sufficient sample size, the results will carry significant weight. In this regard, it should be noted that the differences between hypertensives and normotensives in terms of sex, age and nationality were not significant (P < 0.154, P < 0.598 and P < 0.964 respectively). The difference between males and females with regard to the age group was also not significant (P < 0.210). Furthermore, we do not think that there was a bias in patient selection because of the consent issue. The main reason for exclusion of subjects from the study was the presence of other chronic diseases rather than them not consenting to participate.
Regarding the questionnaire versus interview, questionnaires were used for those who could read and write, while the same questions were asked using a semi-structured interview format for illiterate subjects.
We agree the contradiction between the table and the text deserves clarification. Table 1 shows that 34.1% of the low income group had hypertension while only 25.1% of the control group had hypertension, indicating an association between low income and hypertension. Similarly, 41.4% of those in the obese group (BMI ≥ 30 kg/m2) were hypertensives as compared to 25.1% of the non-obese group, suggesting an association between higher BMI and hypertension. In Table 3, the percentage of hypertensive subjects with positive family history of diabetes (19.6%) was significantly higher than the percentage (6.3%) of normotensive subjects with positive family history of diabetes. We believe that the confusion arose due to the fact that the direction of association was not clearly indicated in Table 4 while describing the odds ratio. While the different groupings are given in brackets, we failed to indicate which of the different subgroups the odds ratios referred to. The associations in Table 4 as follows: Odds of having hypertension for obese subjects was 4.29 times when compared to those subjects with BMI < 30 kg/m2; the odds of having hypertension for the low-income group (< 5000 dirhams) was 2.69 times higher than the other group (5000+ dirhams); the odds of having hypertension for subjects with positive family history of diabetes was 2.58 times higher than the subjects with negative family history of diabetes; the odds of having hypertension in those who reported physical inactivity was 1.8 times higher than those who reported being physically active; and those with more than 3 children were 1.67 times at higher risk of having hypertension as compared to those with fewer than 3 children. Thus, there are no contradictions as such in the results.
References
- Sabri S, Bener A, Eapen V, Abu Zeid MSO, Al-Mazrouei AM, Singh J. Some risk factors for hypertension in the United Arab Emirates. Eastern Mediterranean health journal, 2004, 10(4/5):610–9.
- Bhopal R. Concepts of epidemiology. Oxford, Oxford University Press, 2002.
- Schlesselmann JJ. Case–control studies. Design, conduct, analysis. Oxford, Oxford University Press, 1982.
- Armitage P. Statistical methods in medical research. Oxford, Blackwell, 1982.
- Matthews DE, Farrewell VT. Using and understanding medical statistics, 2nd ed. Basel Switzerland, Karger, 1988.
S. Sabri and V. Eapen
Department of Psychiatry, Faculty of Medicine & Health Science
UAE University
Al-Ain, UAE
A. Bener
Department of Medical Statistics and Epidemiology
Hamad General Hospital & Hamad Medical Corporation
Weill Cornell Medical College in Qatar
Doha
Qatar
Correspondence to:
M.S.O. Abu Zeid
Primary Health Care Clinic
Oud Al Thoba
Ministry of Health,
Al-Ain, UAE
M. Al-Mazrouei
Khalifa Hospital
Abu Dhabi Health Authority
Abu Dhabi, UAE
J. Singh
Department of Biological Sciences
University of Central Lancashire
Preston, Lancashire
United Kingdom.
