Magnitude of Diabetes Comorbidity among People Living with HIV: A Systematic Review

Although the clinical relationship between HIV/AIDS and Diabetes is well established, there is a little summarized evidence about the magnitude ofDiabetescomorbidity among people liv ing with HIV.The aim of this study was to summarizeevidence on the incidence and prevalence of Diabetes comorbid ityamong people living with HIV.A systematic review of the incidence and prevalence of Diabetes comorbid ity among people living with HIV was conducted. Studies reporting incidence and/orprevalence of diabetes among people living with HIV were retrieved from Embase and Medline databases. Selection of the studieswas based on both relevance and quality.Study and outcome characteristics, were ext racted using a standardized checklist. Resulted were presented using narrative and graphic summaries. PRISMA checklist was used as a guide for reporting of the review results. A total of 12 studies met the inclusion and quality. In total of 43,296 people liv ing with HIV, 1,144 incident cases of diabetes were identified over 174,574 person-years. The incidence rates of diabetes comorbid ity as reported by these studies ranged from 5.72 to 23.8 per 1000 person-years. Similarly, a total of 1,724 prevalent cases of diabetes were identified among 41,068 people living with HIV. The reported prevalence rates ranged from 2.85% to 14.9%. High level of variab ility in the reported incidence and prevalence rates of diabetes comorb idity was observed. Generally, the findings are not in favour of a significantly increased magnitude of diabetes comorbidity in HIV.


Introduction
While still having adisproportionate burden of infectious diseases, most low and middle income countries are now facing a rapid ly emerg ing burden of Noncommunicab le diseases. This epidemio logical overlap is expected to result in higher rates of co morbid ity [1]. It is generally well understood that infectious diseases increase the risk of no n-c o m m uni c abl e dise as es . Co n ve rs ely , non-communicable diseases also predispose people to infectious diseases [2]. Besides, some infectious diseases increase the risk of getting other infectious diseases resulting in higher coinfection rates [3]. This also holds true for Noncommun icable d iseases. Hence, the interrelat ionships between diseases are so complex that comorb idities exist in a significant magnitude and are worth considering.
There is no standard defin ition o f co mo rbid ity. It may refer to diseases or disorders that exist together with an index disease; or it may refer to the co-occurrence o f t wo or mo re diseases in an ind ividual. Measurement of co mo rbid ity is currently gett ing more attention in med ical research . Data sources that are available to measure como rbidity include med ical records, patient self-report, clinical judgement and administrative databases. Co morbidity can take four major forms in statistical analysis: a confounder, an effect mod ifier, an exposure and an outcome. Most comorbid ity studies involve the development andapplication of co morb idity indices to understand the effect of comorbidities on defined health outcomes [4].
Diabetes como rbidity in HIV infection, the occurrence of Diabetes in an individual living with HIV, is receiv ingmo re concernin line with the emergence of d iabetes as a major disease of public health importance in high HIV prevalent countries. Low and middle income countries which already have high magnitude of HIV are expected to share high burden of Diabetes comorbidity due to the increase in the incidence of Diabetes in these settings [5].
Though more is known about the clinical relat ionship between HIV and Diabetes and the factors associated with that relationship, little is known about the epidemio logy of their co morb idity. W ith the emergence of double mortality burden in low and middle inco me countries, better understanding of the epidemiology of HIV-Diabetes comorbid ity is essential [6,7].Data on the incidence and prevalence of Diabetes among people with HIV are availab le fro m d ifferent studies in different parts of the world. There are also several studies that have reported the effect of antiretroviral treat ment on the risk of developing Diabetes among people living with HIV. However, to the knowledge of the investigators, there is no research effort that summarizes altogether the incidence and prevalence of Diabetes comorbid ityin HIV infection.
Therefore, the aim of this study was to review and summarizethe incidence and prevalenceof Diabetes comorbid ity among people liv ing with HIV andthereby to discuss their imp licat ions on possible integration of HIV and Diabetes programs.

Search Strategy and Study Selection
We searched studies reportingtheincidence and/or prevalence of Diabetes comorbid ity among people living with HIV fro m Embase and Medlineelectronic databases (March 2012). We used the key words of "HIV and Diabetesand incidence" and "HIV and Diabetes and prevalence"in Embase search. We used the MeSH terms "HIV infection" and "Diabetes mellitus" in Medline. The titles and abstracts of the identified studies were screened for elig ibility. The references of the identified reviews were consulted without restriction to identify additional studies.
We reviewed the full texts of all the retrieved articles for assessment ofelig ibility. The inclusion of the art icles in to the review was based on the following eligib ility criteria: 1) Studies reporting primary data (reviews were excluded); 2) studies reporting the prevalence and/or incidence of diabetes 3) studies conducted entirely on people living with HIV; and 4) Studies reporting exp licit study design, sample size, follow up period, and outcome ascertainment.
For practical reasons, one author (TN) conducted the screening and the elig ibility assessment and the other authorshave re-checked the screening and the eligib ility assessment process and outputs.
A total of 22 studies met the eligib ility criteria. Nine of these have reported incidence of diabetes in people living with HIV. Sixteen studies have reported prevalence of diabetes comorbid ity among people living with HIV. Three studies have reported both on the incidence and prevalence of diabetes comorbidity among people living with HIV.

Quality Assessment
Risk of bias in the incidence and prevalence estimates was assessed by using five major criteria (adequacy of sample size, patient selection methods used, diabetes diagnosis methods used, quality assurance schemes applied and analysis conducted to account for the effects of possible confounding factors). Each of these criteria were scored out of four points (4=Acceptable, 3=Adequate, 2=inadequate, 1=not acceptable) by two reviewers. Ten articles with scores less than half of the total score (<10 points) were excluded.
The main reason for exclusion was inadequate sample size used for the estimation of incidence and prevalence. The final 12 studies were included in this review.

Data Extraction
We have extracted two categories of data fro m the identified studies using pre-testedchecklists. The first set of variables extracted were study characteristics including name of first author, year of publication, study design, follow up period (if applicable), samp le size, and study population characteristics (age, sex, Body Mass Index (BMI), treat ment status). The second set of variables was outcome characteristics which includebaseline prevalence (if applicable), incidence rates, prevalence rates and values for measures of association.
Two rev iewers independently ext racted data fro m the selected studies. Discrepancies were settled through consensus and by rechecking the individual study reports.

Synthesis Methods
For the included studies confidence intervals for the incidence and prevalence rates were recalculated using poison exact method in using Stata 12. As Meta-analysis of the incidence and prevalence rates was precluded by heterogeneity of the studies, we have summarized the findings of the review using narrative and graphic summaries.

Study Characteristics
Of the 12 studies included in this systematic rev iew, five have reported incidence and seven have reported on the prevalence (one study has reported both prevalence and incidence rates of diabetes). The search and selection of the studies is shown using the PRISMA flo w diagram ( Figure 1).
All the included studies have sample size of greater than 500 and were published between 2005 and 2012. Detailed study characteristics and outcome characteristics including the prevalence and incidence of diabetes (with 95% confidence interval)fro m the included studies are presented in Table 1.

Inci dence of Diabetes Comor bi di ty
In the five incidencestudies, a total of 43,296 people liv ing with HIV were follo wed over a period of 174,574 person-years. During this follow up period a total of 1,144 incident cases of diabetes were identified. The incidence rates of diabetes comorbidity as reported by these studies ranged fro m 5.72 to 23.8 per 1000 person-years. The review results of individual incidence studies are described as follows.  person-years of follo w up.In this study, the incidence of Diabetes has increased with cumulative exposure to combination antiretroviral treat ment. About three-fourth of the study subjects enrolled in to this study were men [8].
In the SwissHIV cohort study, 123 of 6, 513 persons included in the follow up developed diabetes during 27,798 person-years of follow-up(mean duration o f follow up was 4.3 years). This was equivalent to an incidence rate of 4.42 cases per 1000 person-years. This study has reported an increased incidence rate ratio among male study subjects and among those who use protease inhibitors and some nucleoside reverse transcriptase inhibitors (NRTI).The med ian age and the median BMI among the study subjects included in this cohort study were 38 years and 22.5 kg/ m2 respectively. Nearly 70% of the study subjects in this study were men [9].
In the Women's Interagency HIV study conducted in six cities of United States, Diabetes was diagnosed in 116 HIV-infected (n=1524) and 36 HIV-uninfected (n=564) wo men over 6, 802 person-years (4962 for HIV infected and 1840 for uninfected). The incidence of d iabetes among HIV infected was about 2.33 cases per 100 person years. HIV-infected wo men who were not taking antiretroviral therapy had a diabetes incidence rate of 1.53 per 100 person-years. Women who were takinghighly active anti-retroviral treat ment containing a protease inhibitor had a rate of 2.50 per 100 person-years and those reporting non-protease inhibitor-containing HAART had a rate of 2.89 per 100 person-years. The med ian age among the HIV positive women was 39.2 years wh ile the med ian BM I was 26.8 kg/m 2 . In this study, cumulative exposure to NRTI was associated with increased incidence of Diabetes [10].

In a cohort study of 1046 HIV-infected patients in 47
French clinical sites, representing 7,846 person-years of follow-up, 111 patients developed diabetes. The incidence rate was 14.1 per 1,000 person-years of fo llo w up. The incidence rate in th is study was 14.6 in men and 12.6 in wo men [11].
In another study conducted in Taiwan among 824 HIV-infected patients attended fro m 1993 to 2006 at National Taiwan University Hospital, 50 incident cases of Diabetes were diagnosed, resulting in an incidence of 13.1 cases per 1,000 person-years of follow-up. In this study, family history, exposure to zidivudine and current use of protease inhibitors were found to be the major risk factors for development of Diabetes [12]. The distribution of the incidence rates and their 95% confidence intervals are shown in the Figure 2.

Prevalence of Diabetes Comor bi dity
In the seven studies that have reported prevalence of diabetes, 41,068 people living with HIV were studied. Among these population, 1,724 prevalent cases of diabetes were identified. The reported prevalence rates ranged fro m 2.85% in the D:A: D baseline prevalence to 14.9% in the baseline prevalence of veterans ageing cohort study.
In the Data Collection on Adverse Events of Anti-HIV Drugs Study, 952 o f the 33,389 patients had diagnosis of diabetes at index visit. Th is makes the baseline prevalence of diabetes to be 2.85% [8].In the veterans ageing cohort study of 3,227 HIV-infected individuals, the baseline prevalence of diabetes was 14.9% in the HIV in fected as compared to 21.4% in the HIV unin fected group. Most of this difference was attributed to the differences in body mass index [13]. In a study of comparison of factors associated with prevalent diabetes mellitus among HIV-infected antiretroviral naive indiv iduals enrolled in to Co mmunity Programs for Clinical Research on AIDS (CPCRA) clinical trials andadults enrolled in to the National Health and Nutrit ional Examination Survey (NHANES), the prevalence of self-reported Diabetes in the CPCRA clinical trials and the NHA NES cohorts was 3.3% and 4.8%, respectively [14].
In a study of the prevalence of diabetes among 610 HIV positive asymptomatic patients in Botswana, the prevalence of diabetes was found to be 4.6% as compared to 7.2% for the general population in Botswana. A mong those above 40 years of age, the prevalence was 10.8%. Older age and higher BM I were associated with higher diabetes prevalence in this study [15].
In a study conducted on 755 HIV positive individuals attending tertiary care hospital for routine clinical and laboratory follo w-up between July and September 2009 in Italy, the prevalence of diabetes was found to be 4.5%. A longer exposure to antiretroviral therapy and a diagnosis of lipodystrophy syndrome were significantly associated with higher levels of d iabetes prevalence [16].
In a study of 643 older men, the prevalence of diabetes based on Oral Glucose Tolerance Test and among HIV infected anti-retroviral naïve indiv iduals was 11% as compared to 8% in HIV uninfected indiv iduals. In this study, age and Ethnicity were associated with abnormal Oral Glucose Tolerance Test (OGTT) results [17].
In the Multi-centre AIDS Cohort study, 68 (11.97%) of the 568 HIV-infected men had prevalent Diabetesat the baseline visit. The prevalence of diabetes was 14% among those 411 HIV positive men who were using antiretroviral treatment. A mong 157 HIV positive men who were not using highly active anti-retroviral treat ment, 11 have diabetes and the prevalence of Diabetes at index visit among this group was about 7%[18].

Discussion
The incidence of Diabetes comorbidity among people liv ing with HIV generally looks to be similar tothat of the general population. But this incidence has shown higher variability between studies with higher person years of follow up and those studies with relatively lo werperson years of follow up. Studies with higher person years of follow up seem to have a lower incidence of d iabetes comorbidity than those with lower person years of follow up. The D: A : D study and the Swiss cohort study [8,9] which have very high person years of fo llo w up have reported incidence rates which are lo wer than all other reported values.
The prevalences of Diabetes co morbidity, wh ich were outputs of cross-sectional studies in most cases, have also shown wider variability among the included studies. The prevalence of Diabetes in the respective general population, the socio-demographic pro files of the study subjects and duration on treatment may be the main factors related with this variability.
The relat ively higher level of incidence and prevalence of diabetes among people living with HIV, and particu larly among those who are receiving anti-retroviral treat ment, warrants the screening of PLHIV for hyperglycaemia both at time of enrolment and during follo w up period of HIV treatment. Whether this screening should be universally implemented for all people living with HIV or selectively for those with other risk factors needs further investigation. Screening during follow up periods while a person is on HIV treatment should also be the main stay in the prevention and control of d iabetes among people living with HIV. Thus, With regard to the limitations of this study, aggregation of the incidence and prevalence of diabetes comorbid ity among people living with HIV in to a single summary figure was not possible due to the heterogeneity of the included studies. This heterogeneity has also increased the variability among the reported incidence and prevalence rates. Another limitat ion of this study is that it was not possible to present the disaggregated incidence and prevalence rates for PLHIV taking ART and those not taking ART as only a few of studies have reported disaggregated values. Lastly, diabetes incidence and prevalence rates for the general population fro m wh ich the study subjects were drawn were not availab le for co mparison.
In conclusion, this review has summarized the incidence and prevalence of diabetes comorbid ity among people living with HIV. High variab ility in the reported incidence and prevalence rates were observed among the studies. Co mparison of diabetes incidence and prevalence between HIV infected and uninfected people is also not consistent among studies. Diabetes prevalence in the general population, Sample sizes, duration of follow up and year of the study may be the possible factors associated with this variability. The study findings, in general, are not in favour of a significantly increased magnitude of diabetes comorbid ity among people living with HIV. However, given the lower average age of the study participants, regular monitoring and screening of blood sugar levels of people liv ing with HIV are reco mmended.
Fundi ng: None Conflict of interest: None declared Ethical approval : Not required