Diabetes Mellitus is Associated with Increased Death Rates Among HIV-Infected Patients in Rio de Janeiro,Brazil

Abstract

Diabetes mellitus (DM) is a major cause of morbidity worldwide and a known factor leading to increased risk of death, especially in conjunction with other risk factors. In this study, we evaluated the prevalence of DM among HIV-infected patients and its association with overall mortality. All HIV-infected patients 18 years or older who were followed in the Instituto Nacional de Infectologia Evandro Chagas (INI) cohort from January 1991 to December 2011 were included. Time-updated covariables included DM status, calendar year, combination antiretroviral therapy (cART), and CD4 cell counts. Fixed demographic covariables included gender and age at entry. Poisson models were used to calculate mortality rate ratios (RR) with robust variances. Among the 4,871 patients included, 1,192 (24.4%) died (mortality rate = 4.72/100 person-years [PY]; 95% confidence interval [CI] = 4.46–5.00). Death rates were significantly higher among those presenting with DM compared with those who did not (6.16/100 vs. 4.61/100 PY, respectively. p = 0.001). In the final model, DM was significantly associated with mortality (RR = 1.74; 95% CI = 1.57–1.94; p < 0.001). When the analysis was restricted to those on cART or the period post-1996, the association between DM and mortality was even stronger (RR = 2.17; 95% CI = 1.91–2.46; p < 0.001 and RR = 1.95; 95% CI = 1.75–2.18; p < 0.001, respectively). Among the major groups of cause of deaths (CODs), the proportion of AIDS-related conditions in patients with DM was lower (74.27% vs. 58.93%, respectively; p < 0.001); whereas in non-AIDS-related conditions, nonimmunodeficiency-related causes (22.44% vs. 34.82%, respectively; p = 0.004) were more common in patients with DM. In conclusion, DM was associated with increased mortality rates even after controlling for HIV-related variables associated to this outcome. Differences in the underlying CODs were identified, reinforcing the necessity to assess and treat comorbidities such as DM in HIV-infected patients.

Introduction

Diabetes mellitus (DM) is a highly prevalent cause of morbidity worldwide, and it operates along a diversity of risk factors to determine the underlying macro- and microvascular dysfunctions seen in cardiovascular diseases (CVDs), which collectively can lead to death.¹ DM-associated mortality is also commonly caused by cerebrovascular disease, peripheral arterial disease, and renal failure, all of which share similar traditional risk factors. So far, investigations have shown that specific antiretroviral regimens have been linked to insulin resistance, dyslipidemia, and body fat redistribution, leading to an increased risk of ischemic heart disease.² Hence, DM occurrence and its association with metabolic syndrome, a cluster of risk factors to CVDs, is becoming highly prevalent among people living with HIV/AIDS (PLWHA) from industrialized and developing countries³ as well. Consequently, CVDs have become a major cause of death (COD) among PLWHA⁴ in the combination antiretroviral therapy (cART) era.

The implementation and scaling of combined antiretroviral therapy (cART) prompted a marked decrease in AIDS-related causes of death and an improvement in survival among HIV-infected individuals, leading to increased awareness in respect to deaths caused by illnesses that were not previously considered AIDS related. The emergence of comorbidities such as non-AIDS-defining malignancies, liver disease, CVD, and DM has been increasingly reported worldwide, irrespective of the development level.^5

–9 Associations linking HIV, AIDS treatment, and apparent increases in so-called “non-AIDS-related” causes of death clearly need more investigation. In the past, we assumed that most cases of “AIDS-related” mortality would be the consequence of immunosuppression, but more recently, it has become clear that AIDS-related or treatment-related metabolic deregulation is also important.¹⁰ Brazil constitutes a unique setting, since it is a large low-income country where cART has become available in a free and universal manner since 1996. We have previously reported the emergence of DM and CVD as COD^6,11 among HIV-infected individuals in Brazil. However, only a few studies have investigated the consequences of diabetes in large cohorts of PLWHA.

In this article, we aimed at evaluating the prevalence of DM in a large cohort of PLWHA in Rio de Janeiro and at assessing the relationship between DM and all-cause mortality.

Methods

Cohort

This study was conducted at the Instituto Nacional de Infectologia Evandro Chagas (INI), located at Oswaldo Cruz Foundation (Fiocruz), Rio de Janeiro State, Brazil. INI is a national reference center for infectious diseases and has been a reference center for care, research, and training related to HIV/AIDS since 1986. An observational, longitudinal, and clinical database—INI cohort, formerly known as the IPEC cohort—is maintained, with PLWHA receiving primary, specialty, and tertiary HIV care at its facilities. Clinical care, laboratory tests, and hospitalization data are all available in electronic and paper records. Viral load and CD4 cell counts are measured two to three times a year as a part of regular care of patients. The CD4/CD8 lymphocyte count is measured by flow cytometry (Becton-Dickinson, USA). Quantitation of HIV RNA is performed using nucleic acid sequence-based amplification methodology (NASBA; NucliSens HIV-1 QT Test, Organon Teknika, Bostel, The Netherlands), with a minimum detection limit of 80 copies/ml. Antiretroviral treatment follows the Ministry of Health recommendations for antiretroviral treatment of adults and adolescents, which is periodically reviewed by a panel of experts in the field. Prophylaxis and treatment of opportunistic infections also follow the recommendations of the Brazilian Ministry of Health.¹²

Data are updated every 6 months using outpatient and inpatient clinical documentation and laboratory testing results comprising measurements of lipids, fasting glucose, and glycated hemoglobin. Prescription of antiretroviral therapy is documented by the medical provider and support staff in the clinical records. Trained abstractors record all information into standardized forms, which are processed into a central database. Further details on the HIV/AIDS cohort can be found elsewhere.¹³ All PLWHA who were followed in the INI cohort who were older than 18 years and had a minimum follow-up of 60 days, from January 1991 to December 2011, were included.

Outcome of interest and operational definitions

Information on the vital status—the outcome of interest—was abstracted from clinical charts. Records of the PLWHA from the cohort were linked with the Rio de Janeiro Mortality Registry using a previously validated linkage algorithm and allowing for specific errors in names and dates of birth. On validation, the algorithm has proved to be highly sensitive and specific in retrieving information on the vital status for those considered lost to follow-up.¹⁴

DM was determined and further assessed from medical charts every 6 months whenever subjects had fasting glucose ≥126 mg/dl or glycated hemoglobin (A1C) >6.5 or clinical history of DM or treatment with oral hypoglycemic agents or insulin. DM occurrence has been defined as cases reported at any time of the follow-up as a chronic non-AIDS-related condition.

Given the incurable trait of such a condition, we considered DM reporting as prevalent cases to include incident cases as well as participants who had DM at baseline examinations.

AIDS diagnosis at baseline was defined by the occurrence of an AIDS-defining illness or CD4 cell count <200/μl at or before cohort enrolment (CDC AIDS case definition¹⁵). Race/ethnicity was based on the provider's report and was categorized into white and other. Schooling was self-reported and categorized according to the number of years in school into three groups: <6 years (low formal education), 6–9 years, and >9 years. Age at enrolment was calculated as the difference between date of birth and date of enrolment and was categorized as 18–30, 31–40, 41–50, 51–60, and >60.

Baseline CD4 cell count was defined within a window of 180 days of enrolment, and, if more than one measurement was available, the value closest to the date of enrolment was chosen and categorized as follows: ≤100, 101–200, 201–350, 351–500, and >500.¹⁶ Individuals were considered to be on cART if they were in use of at least three antiretroviral drugs: two nucleoside reverse transcriptase inhibitors and either one protease inhibitor or one non-nucleoside reverse transcriptase inhibitor; they were categorized into those who, at any given point in time, used cART for more than 12 weeks, and those who never used cART (never used or used it for 12 weeks or less). HIV exposure categories were self-reported and categorized hierarchically into injection drug use (IDU, irrespective of sexual practices), men who have sex with men (MSM, including bisexuals), heterosexual, and other/unknown.

All death events that occurred in the study period were classified according to coding causes of death in HIV (CoDe) methodology (http://www.cphiv.dk/DAD/). CoDe is a standardized method for specifying the underlying COD in HIV-positive individuals. It has been designed for clinical trials and epidemiologic studies and includes detailed data collection on the causes of death and contributing factors and a centralized review process of information.¹³ Groups of causes of death were defined according to the categories presented in the CoDe review form.

Statistical methods

Descriptive statistics included the median and interquartile range for continuous variables and frequency distributions for categorical variables. A table was created to compare the characteristics between alive and dead patients in the cohort at the end of follow-up as well as characteristics of diabetic (ever) versus nondiabetic subjects. Differences were assessed with the nonparametric Mann–Whitney test for continuous variables and Chi-squared tests for categorical variables. Mortality rates were expressed as number of deaths per 100 person-years (PY), with 95% confidence intervals (95% CIs) calculated using the exact Poisson method. Multivariate analyses with Poisson models were used to calculate rate ratios (RR), with robust variances for the association between DM and mortality in the entire cohort period and after restricting the analysis to the period post-1996, as well as to subjects ever on cART. For descriptive statistics, all variables were considered fixed; however, for the Poisson models, DM status, calendar year, cART use, and CD4 cell counts were treated as categorical, time-updated covariables whereas all other covariables were fixed.

Chi-squared or Fisher's exact test, when appropriate, summarizes differences in the observed proportions of AIDS-related mortality according to DM status. p < .05 were considered statistically significant. All analyses were done in the R environment version 3.1.1.¹⁷

Ethical approval

This study was conducted according to the Declaration of Helsinki principles. INI PLWHA cohort was approved by the Evandro Chagas Clinical Research Ethics Committee (043/2010, 0032.0.009.000-10). All participants provided written informed consent.

Results

The study comprised 4,871 participants in the cohort, which contributed to 24,165 PY of follow-up. Among these, 1,192 (24.4%) died (death rate = 4.72/100 PY; 95% CI = 4.46–5.00) and 498 had a DM diagnosis, yielding to a DM prevalence of 10.2%. Overall median age was 34.8 years; in patients who were alive, age was slightly lower compared with those who were dead (median of 34.4 and 35.8, respectively; p < .001). The proportion of men was 67.5%, and 47.4% had low formal education. In comparison to alive patients, dead patients were more likely to be men, to have low formal schooling, to present lower CD4 cell counts, and to have an AIDS diagnosis at the baseline and they were less likely to have ever experienced cART (Table 1). Patients with DM were twice as likely to have dyslipidemia when compared with patients without DM (50.4% vs. 21.9%; p < 0.001). In addition, patients with DM were older (41.2% in DM vs. 34.1 in non-DM, respectively; p < 0.001) and more likely to belong the heterosexual risk category group (51% in DM vs. 45.7% in non-DM; p = 0.042). However, despite them being more likely to use cART (91.6% in patients with DM vs. 70.6% in patients without DM; p < 0.001), CDC criteria for AIDS (59% in patients with DM vs. 51.8% in patients without DM; p = 0.003) and pulmonary tuberculosis (14.75 in patients with DM vs. 9.8 in patients without DM; p < 0.001) were more frequent in DM participants (Table 2). Crude analysis comparing the mortality rates of diabetics with nondiabetics showed that death rates were significantly higher among patients with DM (6.16/100 PY vs. 4.61/100 PY among those without DM, p = 0.001).

Table 1.

Descriptive Statistics and Comparisons Between Alive and Dead Patients at the End of Follow-Up, 1991–2011

	Alive	Dead	Total	Missing	p^a
Total	3,679	1,192	4871	0
DM				0	.49
No	3,296 (89.6)	1,077 (90.4)	4,373 (89.8)
Yes	383 (10.4)	115 (9.6)	498 (10.2)
Dyslipidemia^b				0	<.001
No	2,635 (71.6)	1,026 (86.1)	3,661 (75.2)
Yes	1,044 (28.4)	166 (13.9)	1,210 (24.8)
Gender				0	<.001
Female	1,266 (34.4)	319 (26.8)	1,585 (32.5)
Male	2,413 (65.6)	873 (73.2)	3,286 (67.5)
Schooling				2.52	<.001
<6 years	1,810 (5.1)	439 (38.7)	2,249 (47.4)
6 to 9 years	973 (26.9)	392 (34.6)	1,365 (28.7)
9 years+	832 (23)	302 (26.7)	1,134 (23.9)
Ethnicity				0.02	.33
White	1,944 (53.1)	649 (54.7)	2,593 (53.5)
Other	1,720 (46.9)	437 (45.3)	2,257 (46.5)
Age (years)				0	<.001
Median (IQR)	34.4 (28.1–41.7)	35.8 (29.1–43.5)	34.8 (28.4–42.1)
Age groups				0	<.001
18–30	1,213 (33)	335 (28.1)	1,548 (31.8)
30–40	1,348 (36.6)	435 (36.5)	1,783 (36.6)
40–50	812 (22.1)	273 (22.9)	1,085 (22.3)
50–60	248 (6.7)	115 (9.6)	363 (7.5)
>60	58 (1.6)	34 (2.9)	92 (1.9)
CD4 counts (cells/mm³)				26.42	<.001
Median (IQR)	326 (142–542)	174 (54.5–362)	300 (117–520)
CD4 categories				26.42	<.001
0–100	591 (19.6)	207 (36.8)	798 (22.3)
100–200	398 (13.2)	100 (17.8)	498 (13.9)
200–350	605 (20)	107 (19)	712 (19.9)
350–500	554 (18.3)	62 (11)	616 (17.2)
>500	873 (28.9)	87 (15.5)	960 (26.8)
Ever on cART^c				0	<.001
No	679 (18.5)	650 (54.5)	1,329 (27.3)
Yes	3,000 (81.5)	542 (45.5)	3,542 (72.7)
Risk category				0	<.001
Heterosexual	1,773 (48.2)	480 (40.3)	2,253 (46.3)
IDU	62 (1.7)	61 (5.1)	123 (2.5)
MSM	1,393 (37.9)	441 (37)	1,834 (37.7)
Others/unknown	451 (12.3)	210 (37.6)	661 (13.6)
CDC/AIDS case definition				0
No	2,034 (55.3)	278 (23.7)	2,312 (47.5)		<.001
Yes	1,645 (44.7)	914 (76.7)	2,559 (52.5)

Columns provide number (percentage), unless otherwise specified.

Chi-squared test.

Defined as LDL >159 and/or HDL <40.

Defined as use of cART for more than 12 weeks.

cART, combination antiretroviral therapy; CDC, Centers for Disease Control and Prevention; DM, diabetes mellitus; HDL, high-density lipoprotein; IDU, injection drug use; IQR, interquartile range; LDL, low-density lipoprotein; MSM, men who have sex with men.

Table 2.

Descriptive Statistics and Comparisons Between Diabetic and Nondiabetic Patients, 1991–2011

	Nondiabetic	Diabetic	Total	Missing	p^a
Total	4373	498	4871	0
Death				0	.484
No	3,296 (75.4)	383 (76.9)	3,679 (75.5)
Yes	1,077 (24.6)	115 (23.1)	1,192 (24.5)
Dyslipidemia^b				0	<.001
No	3,414 (78.1)	247 (49.6)	3,661 (75.2)
Yes	959 (21.9)	251 (50.4)	1,210 (24.8)
Gender				0	.209
Female	1,410 (32.2)	175 (35.1)	1,585 (32.5)
Male	2,963 (67.8)	323 (64.9)	3,286 (67.5)
Schooling				2.52	.075
<6 years	2,034 (47.8)	215 (43.4)	2,249 (47.4)
6 to 9 years	1,017 (23.9)	117 (23.6)	1,134 (23.9)
9 years+	1,202 (28.3)	163 (32.9)	1,365 (28.7)
Ethnicity				0.02	.99
White	2,327 (53.5)	266 (53.4)	2,593 (53.5)
Others	2,025 (46.5)	232 (46.6)	2,257 (46.5)
Age (years)				0	<.001
Median (IQR)	34.1 (28–41.2)	41.2 (34–46.8)	34.8 (28.4–42.1)
Age at diagnosis				0	<.001
Median (IQR)	32.2 (26.5–39.3)	37.7 (31–44.7)	32.8 (26.9–40)
Age groups				0	<.001
18–30	1,480 (33.8)	68 (13.7)	1,548 (31.8)
30–40	1,618 (37)	165 (33.1)	1,783 (36.6)
40–50	913 (20.9)	172 (34.5)	1,085 (22.3)
50–60	292 (6.7)	71 (14.3)	363 (7.5)
>60	70 (1.6)	22 (4.4)	92 (1.9)
Age at diagnosis groups				0	<.001
0–18	85 (1.9)	2 (0.4)	87 (1.8)
18–30	1,674 (38.3)	105 (21.1)	1,779 (36.5)
30–40	1,609 (36.8)	172 (34.5)	1,781 (36.6)
40–50	728 (16.6)	147 (29.5)	875 (18)
50–60	220 (5)	60 (12)	280 (5.7)
>60	57 (1.3)	12 (2.4)	69 (1.4)
Risk category				0	.042
Heterosexual	1,999 (45.7)	254 (51)	2,253 (46.3)
IDU	114 (2.6)	9 (1.8)	123 (2.5)
MSM	1,672 (38.2)	162 (32.5)	1,834 (37.7)	0
Others/unknown	588 (13.4)	73 (14.7)	661 (13.6)
IDU				0	.354
Others	4,259 (97.4)	489 (98.2)	4,748 (97.5)
IDU	114 (2.6)	9 (1.8)	123 (2.5)
CD4 counts (cells/mm³)				26.42	.355
Median (IQR)	301.5 (120–520.8)	288 (101–516.8)	300 (117–520)
CD4 categories				26.42	.242
0–100	695 (21.9)	103 (24.9)	798 (22.3)
100–200	436 (13.8)	62 (15)	498 (13.9)
200–350	646 (20.4)	66 (15.9)	712 (19.9)
350–500	544 (17.2)	72 (17.4)	616 (17.2)
>500	849 (26.8)	111 (26.8)	960 (26.8)
Ever on cART^c				0	<.001
No	1,287 (29.4)	42 (8.4)	1,329 (27.3)
Yes	3,086 (70.6)	456 (91.6)	3,542 (72.7)
CDC/AIDS case definition				0	.003
No	2,108 (48.2)	204 (41)	2,312 (47.5)
Yes	2,265 (51.8)	294 (59)	2,559 (52.5)
Pulmonary tuberculosis				0	<.001
No	3,945 (90.2)	425 (85.3)	4,370 (89.7)
Yes	428 (9.8)	73 (14.7)	501 (10.3)

Columns provide number (percentage), unless otherwise specified.

Chi-squared and Mann–Whitney tests.

Defined as LDL >159 and/or HDL <40.

Defined as use of cART for more than 12 weeks.

In the final Poisson model, DM was significantly associated with mortality (RR = 1.74; 95% CI = 1.57–1.94; p < 0.001). Male sex, older age, CD4 counts, cART use, and higher education were also associated with mortality. Furthermore, death rates declined significantly over the cohort enrollment period (Table 3). The association was even stronger when the analysis was restricted to those who used cART (RR = 2.27; 95% CI = 1.78–2.89; p < 0.001) or the period post-1996 (RR = 2.11; 95% CI = 1.70–2.62; p < 0.001) (Table 4).

Table 3.

Final Poisson Model ^a with Estimated Rate Ratios and 95% Confidence Interval for the Effect of Diabetes Mellitus on Mortality

Variable	RR (95% CI)	p^b
DM = yes	1.74 (1.57–1.94)	<.001
Male gender	1.30 (1.21–1.40)	<.001
Age group^c
20–30	1.04 (0.83–1.30)	.18
30–40	1.16 (0.93–1.43)	.001
40–50	1.45 (1.16–1.81)	<.001
50–60	2.15 (1.70–2.71)	<.001
>60	2.60 (1.92–3.53)	<.001
CD4 cell count^d
Unknown	0.36 (0.33–0.39)	<.001
>100	0.12 (0.11–0.13)	<.001
cART = yes	0.33 (0.30–0.37)	<.001
>9 years of schooling^e	0.76 (0.71–0.81)	<.001
CDC/AIDS case definition = yes	2.29 (2.12–2.48)	<.001
Year^f
1992	0.92 (0.74–1.14)	.44
1993	0.92 (0.76–1.12)	.41
1994	0.93 (0.77–1.13)	.47
1995	0.75 (0.61–0.91)	.005
1996	0.49 (0.38–0.62)	<.001
1997	0.71 (0.57–0.90)	.004
1998	0.62 (0.48–0.79)	<.001
1999	0.67 (0.53–0.85)	.001
2000	0.61 (0.48–0.78)	<.001
2001	0.75 (0.59–0.94)	.01
2002	0.47 (0.36–0.60)	<.001
2003	0.67 (0.54–0.84)	.001
2004	0.94 (0.75–1.17)	.56
2005	0.92 (0.74–1.14)	.44
2006	0.63 (0.51–0.79)	<.001
2007	0.78 (0.63–0.96)	.02
2008	0.68 (0.55–0.85)	.001
2009	0.72 (0.58–0.90)	.004
2010	0.65 (0.52–0.80)	<.001
2011	0.61 (0.43–0.86)	.02

Data of all cohort participants.

DM status, calendar year, cART use, and CD4 cell counts were treated as time-updated covariables; whereas all other covariables were fixed.

From January 1991 to December 2011.

Obtained with the robust variance model.

Reference category = 0–20 years.

Reference category ≤100 cells/mm³.

Reference category <9 years of schooling.

Reference category = 1991.

CDC, Centers for Disease Control and Prevention; CI, confidence interval; RR, rate ratios.

Table 4.

Poisson Model with Estimated Rate Ratios and 95% Confidence Interval for the Effect of Diabetes Mellitus on Mortality of Cohort Members, Restricted for Post-1996 Period and Combination Antiretroviral Therapy Use

	Post-1996 ^a		Ever on cART ^b
Variable	RR (95% CI) ^c	p^c	RR (95% CI) ^c	p^c
DM = yes	1.95 (1.75–2.18)	<.001	2.17 (1.91–2.46)	<.001
Male gender	1.30 (1.20–1.42)	<.001	1.23 (1.10–1.37)	<.001
Age group^d
20–30	0.90 (0.71–1.14)	.37	0.49 (0.37–0.64)	<.001
30–40	0.94 (0.75–1.18)	.60	0.49 (0.38–0.64)	<.001
40–50	1.28 (1.01–1.62)	.04	0.61 (0.47–0.80)	<.001
50–60	1.87 (1.45–2.40)	<.001	0.84 (0.64–1.11)	<.001
>60	2.73 (1.98–3.76)	<.001	1.51 (1.05–2.16)	<.001
CD4 cell count^e
Unknown	0.30 (0.27–0.34)	<.001	0.28 (0.25–0.33)	<.001
>100	0.11 (0.10–0.12)	<.001	0.16 (0.14–0.17)	<.001
>9 years of schooling^f	0.68 (0.63–0.74)	<.001	0.63 (0.56–0.69)	<.001
CDC/AIDS case definition	2.32 (2.11–2.55)	<.001	2.83 (2.47–3.25)	<.001
cART = yes	0.32 (0.29–0.35)	<.001

DM status, calendar year, cART use, and CD4 cell counts were treated as time-updated covariables, whereas all other covariables were fixed.

From January 1997 to December 2011.

Used cART for more than 12 weeks.

Obtained with robust variance model, adjusted for calendar year.

Reference category = 0–20 years.

Reference category ≤100 cells/mm³.

Reference category <9 years of schooling.

COD was ascertained for 98.7% of the patients (1,177 out of 1,192). Among the major groups of CODs, AIDS-related conditions accounted for the bulk of deaths regardless of DM status (72.81%); the proportion of AIDS-related conditions in patients with DM was lower when compared with the proportion among those without DM (74.27% vs. 58.93%, respectively; p < 0.001). However, nonimmunodeficiency-related causes (22.44% vs. 34.82%, p = 0.004) and unknown causes (3.29% vs. 6.25%, p = 0.10) were more common in patients with DM (Table 5). The analyses of deaths by specific causes, respectively, revealed that patients with DM died from AIDS malignancies in a higher proportion (14.29%) compared with patients without DM (5.26%, p < 0.001). Similar results have been ascribed to liver-related diseases (4.46% vs. 1.5%, p = 0.04) and cardiovascular diseases (6.25% vs. 3.95%, p = 0.21), although in the latter data are too small to draw any definitive conclusion.

Table 5.

Causes of Death According to Diabetes Mellitus Status

		DM
CoDe	All deaths by major groups	No	%	Yes	%	p^a
	Total	1,065	100	112	100
01.1, 01.2, 01	AIDS related	791	74.27	66	58.93	<.001
02–30	Non-AIDS related	239	22.44	39	34.82	.004
90 and 92	Unknown	35	3.29	7	6.25	.10

		DM
CoDe	All deaths by specific causes	No	%	Yes	%
	Total	1,065	100	112	100
01.1	AIDS infection	476	44.69	39	34.82	.04
01.2	AIDS malignancy	56	5.26	16	14.29	<.001
01	AIDS not specified	259	24.32	11	9.82	<.001
8, 9, 24	Cardiovascular disease	42	3.94	7	6.25	.21

CoDe	Cardiovascular subcategories	No	%	Yes	%
8	MI or ischemic heart disease	14	1.31	2	1.78	.66
9	Stroke	4	0.37	0	0	.99
24	Heart or vascular	24	2.26	5	4.47	.18
90, 92	Unknown	35	3.29	7	6.25	.10
02	Non-AIDS infection	44	4.13	6	5.36	.46
03	Liver related	16	1.50	5	4.46	.04

CoDe	Liver subcategories	No	%	Yes	%
03.1	HCV	1	0.10	0	0	.99
03.1.1	HCV with cirrhosis	7	0.66	2	1.78	.20
03.1.2	HCV with liver failure	4	0.36	1	0.90	.39
03.2.1	HBV with cirrhosis	2	0.19	0	0	.99
03.2.2	HBV with liver failure	2	0.19	2	1.78	.04
4	Non-AIDS malignancy	37	3.47	4	3.57	.98
5	DM	0	0	4	3.57	<.001
13	Respiratory disease	7	0.66	2	1.79	.20
14	Toxicity	7	0.66	2	1.79	.20
23	Central nervous system disease	5	0.47	2	1.79	.13
6	Pancreatitis	8	0.75	2	1.79	.24
	Other causes^b	8	0.75	2	1.79	.24
10	Digestive system disease	7	0.66	1	0.89	.55
15	Renal failure	8	0.75	1	0.89	.59
25	Respiratory disease	4	0.38	1	0.89	.39
16	Violence	37	3.47	0	0.00	.04
17, 22	Psychiatric	4	0.38	0	0.00	.98
20	Hematological disease	5	0.47	0	0.00	.98

Chi-squared and Fisher's exact tests.

Causes of death for which less than four deaths were observed were grouped in the “Other causes” group (CoDe 11, 12, 19, 27, and 28).

CoDe, coding causes of death in HIV; HBV, hepatitis B virus; HCV, hepatitis C virus; MI, myocardial infarction.

Discussion

In this article, we describe the prevalence of DM and its association with mortality rates among adults living with HIV in Rio de Janeiro, Brazil, a middle-income country providing free and universal antiretroviral therapy since 1996. The study showed a significantly higher mortality rate among HIV patients with DM, even after adjusting for gender, age, CD4, cART use, and schooling. Besides, despite such a low median age, DM was a significant contributor to excess mortality. Moreover, CODs commonly associated with diabetes such as cardiovascular events and renal failure were not the main drivers of the observed mortality differential. The apparent contribution of malignancy and hepatic disease needs to be closely observed.

Because both the causes and timing of death changed dramatically in the cohort during the late 1990s and mortality predictors in subjects who enrolled early may have been quite different than those in subjects who enrolled later, statistical models have been adjusted for calendar year. Furthermore, when the analysis was restricted to the period after 1996, the adjusted rate ratio for the association between DM and mortality was even higher.

These results are consistent with other large cohort studies.^7,9 For instance, data from SMART and SPIRIT combined showed that diabetes is an independent predictor for death after a non-AIDS event.¹⁸ In the D.A.D study, DM was also associated with an increased risk of all causes of mortality considered, with an exception for non-AIDS malignancies.² As PLWHA on cART seems to be at higher risk of insulin resistance and DM compared with HIV-uninfected patients, there is a need to call attention to screen and properly manage diabetes both before and after cART initiation¹⁹ to prevent the occurrence of nonfatal events as well as to increase life expectancy. The adjusted rate ratio was significantly higher among those with DM and even greater when restricting the analyses to those using cART, pointing to an increasing DM prevalence as a result of augmented survival. Moreover, AIDS treatment itself, in addition to older age, may increase the incidence of DM in survivors. Currently, cART coverage in Brazil is up to 80%²⁰; herein, the proportion of patients ever on cART was 72.7%.

The observed prevalence of DM (10.2%) was slightly more elevated than the prevalence estimated for the overall Brazilian population (6.0%),²¹ even though it is comparable to other HIV populations under cART treatment.²² The emergence of DM in PLWHA is likely due to the confluence of factors, including the increasing proportion of overweight and obesity,²³ the growing prevalence of metabolic syndrome,³ HCV coinfection,^24,25 and the metabolic complications of certain classes of antiretroviral, mainly protease, inhibitors.^26,27 However, aging of the individuals living with HIV would lengthen their exposure to the same risk factor that is believed to determine DM in the general population.²⁸

AIDS-related malignancies were more frequent among patients with DM in comparison to patients without DM. Recently, we pointed to Kaposi's sarcoma and non-Hodgkin's lymphomas as the most common COD within the AIDS-malignancy group in the same cohort.¹³ Although DM has been associated to increased risks for a wide range of cancer types^29,30 and classical Kaposi's sarcoma has been strongly and independently associated to DM,³¹ no conclusion has been drawn so far in relation to non-Hodgkin's lymphomas.³⁰

Hepatitis C has been linked both to DM through insulin resistance and to non-Hodgkin's lymphoma. Concurrent DM and chronic hepatitis C increase the rate of progression to liver failure and hepatocellular carcinoma.³² Depending on the prevalence of HCV in this cohort, it could be a partial explanation for the observed COD patterns in Table 5. The combination of diabetes and alcoholism would also contribute to liver deaths.

The higher prevalence of alive patients with DM, relative to those who died, also suggests that DM might accelerate the progression of other disease entities in the setting of HIV, as reported with hepatitis C virus coinfection.³³ This can explain the increased mortality rate associated with DM.

The present study has some important weaknesses, mainly regarding confounding effects. Thus, this article lacks data regarding covariates that are considered well-established mortality risk factors, such as obesity, smoking, alcoholism, hypertension, and hepatitis C serostatus. In addition, we were unable to assess type-1 DM in our study and misclassification could have occurred if diagnoses of DM status were missed by physicians or via the data collection methods employed. Information of regimen adherence and HIV viral load evolution not only might have had an impact on overall mortality but were also unavailable in our database. This would result in an over- or underestimation of risk ratios, but both the comprehensive way in which DM diagnosis was ascertained and the similar magnitude of results to other studies in literature make us expect that these effects were minimal.

To warrant reliability of the study, measures for collecting and processing data were established as follows: standardization of procedures according to the current guidelines, trained staff to monitor and review data records, and the use of a high-sensitivity linkage algorithm to minimize losses to follow-up.

Although it is not feasible to ensure whether the presented data are representative of the Brazilian population with HIV/AIDS, they are very likely to represent those living in large urban centers in Brazil.

Collectively, our data reinforce the need to assess and manage chronic comorbidities, particularly DM in HIV populations in settings where cART use is being expanded through public health interventions so that its benefits can be fully experienced. The importance of endorsing the same preventive measures directed to the general population in the direction of adherence to screening guidelines and glycemic control into the preventive care of HIV-infected patients is instrumental to improve survival and quality of life among such individuals.

Conclusions

DM was significantly associated to increased mortality rates among HIV-infected patients, even after controlling for HIV-related variables associated to this outcome. Moreover, we identified some differences in the underlying COD among patients with DM. Additional studies should investigate the causal pathways that appear to link diabetes with deaths from AIDS-associated malignancies and liver disease in similar populations, so that the most effective intervention opportunities could further be identified.

Footnotes

Acknowledgments

This work has been supported by the Fundação Carlos Chagas Filho de Amparo a Pesquisa do Estado do Rio de Janeiro [E-26/201.471/2014 to AGP]; Conselho Nacional de Desenvolvimento Científico e Tecnológico [407446/2012-5 and 305801/2015-5 to AGP]; Programa Estratégico de Apoio à Pesquisa em Saúde - Fundação Oswaldo Cruz [407446/2012-5 to AGP].

Author Disclosure Statement

No competing financial interests exist.

References

Cheng

, Claggett

, Correia

, et al.: Temporal trends in the population attributable risk for cardiovascular disease: The atherosclerosis risk in communities study. Circulation, 2014; 130:820–828.

Data Collection on Adverse Events of Anti HIV drugs Study Group, Smith

, Sabin

, et al.: Factors associated with specific causes of death amongst HIV-positive individuals in the D:A:D Study. AIDS, 2010; 24:1537–1548.

Paula

, Falcao

, Pacheco

: Metabolic syndrome in HIV-infected individuals: Underlying mechanisms and epidemiological aspects. AIDS Res Ther, 2013; 10:32.

Crabtree-Ramirez

, Del Rio

, Grinsztejn

, Sierra-Madero

: HIV and noncommunicable diseases (NCDs) in Latin America: A call for an integrated and comprehensive response. J Acquir Immune Defic Syndr, 2014; 67 Suppl 1:S96–S98.

Palella

Jr. , Baker

, Moorman

, et al.: Mortality in the highly active antiretroviral therapy era: Changing causes of death and disease in the HIV outpatient study. J Acquir Immune Defic Syndr, 2006; 43:27–34.

Paula

, Schechter

, Tuboi

, et al.: Continuous increase of cardiovascular diseases, diabetes, and non-HIV related cancers as causes of death in HIV-infected individuals in Brazil: An analysis of nationwide data. PLoS One, 2014; 9:e94636.

Hasse

, Ledergerber

, Furrer

, et al.: Morbidity and aging in HIV-infected persons: The Swiss HIV cohort study. Clin Infect Dis, 2011; 53:1130–1139.

Chu

, Umanski

, Blank

, Meissner

, Grossberg

, Selwyn

: Comorbidity-related treatment outcomes among HIV-infected adults in the Bronx, NY. J Urban Health, 2011; 88:507–516.

Lucas

, Chaisson

, Moore

: Survival in an urban HIV-1 clinic in the era of highly active antiretroviral therapy: A 5-year cohort study. J Acquir Immune Defic Syndr, 2003; 33:321–328.

10.

Grinspoon

, Carr

: Cardiovascular risk and body-fat abnormalities in HIV-infected adults. N Engl J Med, 2005; 352:48–62.

11.

Pacheco

, Tuboi

, Faulhaber

, Harrison

, Schechter

: Increase in non-AIDS related conditions as causes of death among HIV-infected individuals in the HAART era in Brazil. PLoS One, 2008; 3:e1531.

12.

F M: Clinical Protocol and Therapeutic Guidelines for Management of HIV Infection in Adults. (2016), www.aids.gov.br/publicacao/2013/. Accessed March 23, 2016 .

13.

Grinsztejn

, Luz

, Pacheco

, et al.: Changing mortality profile among HIV-infected patients in Rio de Janeiro, Brazil: Shifting from AIDS to non-AIDS related conditions in the HAART era. PLoS One, 2013; 8:e59768.

14.

Pacheco

, Saraceni

, Tuboi

, et al.: Validation of a hierarchical deterministic record-linkage algorithm using data from 2 different cohorts of human immunodeficiency virus-infected persons and mortality databases in Brazil. Am J Epidemiol, 2008; 168:1326–1332.

15.

From the Centers for Disease Control and Prevention. 1993 revised classification system for HIV infection and expanded surveillance case definition for AIDS among adolescents and adults. JAMA, 1993; 269:729–730.

16.

Luz

, Grinsztejn

, Velasque

, et al.: Long-term CD4+ cell count in response to combination antiretroviral therapy. PLoS One, 2014; 9:e93039.

17.

R: A language and environment for statistical computing. [computer program]. Version 3.1.1. Vienna, Austria: R Foundation for Statistical Computing, 2014.

18.

Neuhaus

, Angus

, Kowalska

, et al.: Risk of all-cause mortality associated with nonfatal AIDS and serious non-AIDS events among adults infected with HIV. AIDS, 2010; 24:697–706.

19.

Durand

, Sheehy

, Baril

, Lelorier

, Tremblay

: Association between HIV infection, antiretroviral therapy, and risk of acute myocardial infarction: A cohort and nested case-control study using Quebec's public health insurance database. J Acquir Immune Defic Syndr, 2011; 57:245–253.

20.

WHO: Global Update on HIV Treatment 2013: Results, Impact and Opportunities 123. Available at: www.who.int/hiv/pub/progressreports/update2013/en/. Accessed April 15, 2015 .

21.

Shaw

, Sicree

, Zimmet

: Global estimates of the prevalence of diabetes for 2010 and 2030. Diabetes Res Clin Pract, 2010; 87:4–14.

22.

Brown

, Cole

, Li

, et al.: Antiretroviral therapy and the prevalence and incidence of diabetes mellitus in the multicenter AIDS cohort study. Arch Intern Med, 2005; 165:1179–1184.

23.

Miguez-Burbano

, Espinoza

, Cook

, et al.: Alcohol, Brain Derived Neurotrophic Factor and Obesity among People Living with HIV. J AIDS Clin Res, 2013; 4:245.

24.

El-Zayadi

, Anis

: Hepatitis C virus induced insulin resistance impairs response to anti viral therapy. World J Gastroenterol, 2012; 18:212–224.

25.

Negro

: Facts and fictions of HCV and comorbidities: Steatosis, diabetes mellitus, and cardiovascular diseases. J Hepatol, 2014; 61:S69–S78.

26.

, Chen

, Sheng

, et al.: Risk factors for incident diabetes mellitus among HIV-infected patients receiving combination antiretroviral therapy in Taiwan: A case-control study. HIV Med, 2009; 10:302–309.

27.

Justman

, Benning

, Danoff

, et al.: Protease inhibitor use and the incidence of diabetes mellitus in a large cohort of HIV-infected women. J Acquir Immune Defic Syndr, 2003; 32:298–302.

28.

Vance

, Mugavero

, Willig

, Raper

, Saag

: Aging with HIV: A cross-sectional study of comorbidity prevalence and clinical characteristics across decades of life. J Assoc Nurses AIDS Care, 2011; 22:17–25.

29.

Liu

, Hemminki

, Forsti

, Sundquist

, Ji

: Cancer risk in patients with type 2 diabetes mellitus and their relatives. Int J Cancer, 2015.

30.

Szablewski

: Diabetes mellitus: Influences on cancer risk. Diabetes Metab Res Rev, 2014; 30:543–553.

31.

Anderson

, Lauria

, Romano

, et al.: Risk factors for classical Kaposi sarcoma in a population-based case-control study in Sicily. Cancer Epidemiol Biomarkers Prev, 2008; 17:3435–3443.

32.

Dyal

, Aguilar

, Bartos

, et al.: Diabetes mellitus increases risk of hepatocellular carcinoma in chronic hepatitis C virus patients: A systematic review. Dig Dis Sci, 2016; 61:636–645.

33.

Greub

, Ledergerber

, Battegay

, et al.: Clinical progression, survival, and immune recovery during antiretroviral therapy in patients with HIV-1 and hepatitis C virus coinfection: The Swiss HIV Cohort Study. Lancet, 2000; 356:1800–1805.