Introduction
Type 2 Diabetes Mellitus (T2DM) is a preventable non-communicable disease requiring a multifactorial approach to management that incorporates lifestyle change and pharmacotherapy.1 Less than ideal management increases the risk of developing complications and comorbidities such as cardiovascular disease (CVD)2 and various unfavorable social and economic penalties.3 Unfortunately, many hurdles remain for the effective management of T2DM.4 The potential toll that a T2DM diagnosis has on mental health is a major concern.5 Depression is more prevalent among people living with T2DM compared with those who are not6 and increases the risk of premature death.7 Studies suggest that the presence of minor psychological morbidity increases non-adherence to T2DM treatment.8 ,9 Elevated stress levels as a result of becoming aware of a T2DM diagnosis may be an underlying cause of worsening mental health, which could also have negative impacts on overall quality of life such as the levels of contact a person may have with friends and family. With rising awareness of T2DM-related stigma,10–12 it may be that many people receiving a diagnosis avoid contact with (or are shunned by) others due to a fear that they may be blamed for their T2DM status, as has been observed among people with liver and lung-related health problems.13–16
The impact of a T2DM diagnosis on mental health and quality of life has been previously investigated but the vast majority of studies employed cross-sectional designs make it very challenging to infer a causal relationship. This is likely to be a major factor behind the mixture of findings reported, with some meta-analyses indicating a low to modest increase in risk of experiencing depression among people living with T2DM,17 ,18 whereas other studies suggest no change in mental health as a result of T2DM diagnosis19 or instead report depression as a risk factor for T2DM.18 ,20 Longitudinal data affords the opportunity to surmount this challenge, but since diabetes status cannot be randomly assigned, this still leaves the door open for confounding. Personality traits that may be consistent over time, for example, like negative affect, may influence mental health, quality of life, and the levels of social contact a person has with others as well as T2DM risk more generally.21 Resolving these issues is crucial to bolster adherence to pharmacotherapy regimens and participation in lifestyle modification programs, and to underline the need for investments in initiatives seeking to support mental health and to ameliorate the risk of social isolation among people with T2DM.22
The purpose of this study was to use a longitudinal study design to examine for potential impacts of a T2DM diagnosis on mental health, quality of life, and a range of types of social contact.
Data
The 45 and Up Study baseline was collected via self-complete survey (response 18%) between 2006 and 2009.23 The Medicare Australia database (the national provider of universal healthcare in Australia) had been used to randomly sample participants. Follow-up between 2010 and 2011 of the first 100 000 baseline respondents was conducted as part of the Social Economic and Environmental Factors (SEEF) Study, also via self-complete survey. A total of 28 057 men and 32 347 women completed the SEEF follow-up (overall response rate of 60.4%, 3.4±0.95 years follow-up time). Ethical approval for the 45 and Up Study was granted by the University of New South Wales Human Research Ethics Committee (HREC 05035/HREC 10186) and the SEEF Study by the University of Sydney Human Research Ethics Committee (ref no. 10-2009/12187).
Exposure and outcomes variables
The key exposure variable was a diagnosis of T2DM. T2DM status was identified in the baseline and follow-up surveys by responses to the question ‘Has a doctor EVER told you that you have diabetes?’. Participants could report yes or no. Although the question did not specify T2DM, evidence suggests that the majority of new diagnoses among people aged 45 years or older are type 2.24
Six outcome variables were examined. The first was the Kessler Psychological Distress Scale (‘K10’),25 which measures symptoms of psychological distress experienced across 4 weeks prior to a participant's completion of the questionnaire. All 10 questions were measured at baseline and follow-up and included whether a participant had felt tired for no reason, nervous, hopeless, restless, depressed, sad, or worthless. Ranging from scores of 10 to 50, scores of 22 or higher on the K10 denote poorer mental health. In line with previous work, a binary variable was constructed with scores of 22 and over identifying participants at high risk of psychological distress.25–27
The second outcome variable was self-reported quality of life. This was measured in the 45 and Up Study with the question ‘in general, how would you rate your quality of life?’. Participants could respond by ticking either ‘excellent’, ‘very good’, ‘good’, ‘fair’, or ‘poor’. A dichotomous variable was constructed with ‘poor’ contrasted with ‘non-poor’ responses.
The remaining four outcome variables each measured different types of social contact and were taken from the shortened version of the Duke Social Support Index.28 At baseline and follow-up, participants were asked to report the number of times in the past week they had: (1) spent time with friends or family they did not live with; (2) talked to someone (friends, relatives, or others) on the telephone; and (3) attended meetings at social clubs or religious groups. A fourth question required participants to report how many people outside their home, but within 1-hour travel time, they felt close to or could rely on.
Sample and statistical analysis
Sampling was based on two criteria: (1) participants having complete data on all outcome variables at baseline and at follow-up; (2) no doctor diagnosed T2DM reported at baseline. The resulting sample comprised 26 344 individuals. The restriction of the sample to only those who did not have a doctor-diagnosed T2DM status at baseline was implemented in order to focus the analysis strictly on the impact of a recent T2DM diagnosis on the aforementioned outcomes. Accordingly, two types of statistical models were fitted following a description of the sample with cross-tabulations. For the indicators of psychological distress and poor quality of life, logistic regressions were fitted and parameters expressed as ORs and 95% CIs. For the indicators of social contacts, which were integer counts and exhibited over-dispersion (where the variance is greater than the mean), negative binomial regressions with parameters expressed as rate ratios (RR) and 95% CIs were fitted. The T2DM diagnosis variable was added to each model, adjusting for participant age. To address potential confounding, two strategies were employed simultaneously. First, each model was fitted with a fixed effects specification.29 Fixed effects means fitting a unique intercept on every participant, eliminating all time-invariant sources of confounding, measured and unmeasured (eg, negative affect), rendering only longitudinal effects observable.30 Consequently, by focusing parameter estimation on within-person change through time, this restricts the parameter estimate to those participants who experienced a change in T2DM and a change in the outcome variable. This specification does not, however, account for potential sources of confounding that are subject to change over time, such as socioeconomic circumstances. As such, time-varying confounders (in addition to age) were adjusted in each model, including annual household income, economic status (eg, employed, retired, unemployed), and couple status (in a couple vs not). All analyses were conducted in Stata V.12 (StataCorp, College Station, Texas, USA).