Post-traumatic stress disorder and glycaemic control

The International Diabetes Federation (IDF) has noted that good glycaemic control can significantly slow the progression of the disease and reduce the risk of developing complications (IDF, 2006). Recent studies in the US reveal that only 43% of people with diabetes have achieved the American Diabetes Association (ADA) guideline target for HbA_1c of 7% (Saaddine et al, 2002; Kerr et al, 2004; ADA, 2007).

In a systematic review, Cramer (2004) identified that nonadherence to diabetes medications was associated with higher HbA_1c levels. While nonadherence to treatment is a complex, multifaceted phenomenon, recent research has explored the relationship between psychiatric disorders and poor medical compliance. Of particular interest is the relationship between treatment nonadherence and post-traumatic stress disorder (PTSD). PTSD is an anxiety disorder with characteristic symptoms developing after exposure to trauma – which can be the diagnosis of a life-threatening disease (Table 1; American Psychiatric Association [APA], 1995). Epidemiological studies indicate that PTSD is a highly prevalent disorder affecting nearly 8 million adults and is twice as prevalent in women compared to men (National Institute of Mental Health, 2006).

Study aims
Few studies have investigated the relationship between PTSD and either type 1 or 2 diabetes. No studies examining the relationship of PTSD, diabetes and glycaemic control in women were identified by the authors. The current study addresses this gap in the literature. It was hypothesised that among a primary care sample of women with diabetes, those with PTSD symptoms would have lower treatment regimen adherence, demonstrated by a higher HbA_1c, compared with those without PTSD symptoms and that this association would not be explained by other sociodemographic variables.

Materials and methods
Study design
A cross-sectional, correlational design was utilised to determine the association of current PTSD symptoms with HbA_1c levels obtained within the past 6 months. Approval was obtained from the Joint Committee on Clinical Investigation (JCCI) at Johns Hopkins Medicine Institutional Review Board.

Participants and procedures
A purposive sample (where the sample is selected by the researcher subjectively) of 20 women with diabetes on a low income without medical insurance was taken from a larger study on the effects of partner violence on immune function conducted at a primary care clinic for the uninsured in Baltimore, MD (Woods et al, 2005). In the original study, exclusion criteria were as follows:

• age <18years or >60years
• pregnant or lactating
• presenting any current active infections, malignancies or autoimmune disease
• use of street drugs
• chronic excessive alcohol intake
• use of corticosteroids.

Obesity and diabetes were not exclusionary owing to the high prevalence of these conditions in urban Baltimore. The purposive sample were similar to the urban population of Baltimore in terms of race and obesity and similar in all aspects to the primary clinic population: participants were predominantly middle aged (mean age: 49.3 years; SD: 9.1 years); African–American (85%); unmarried (30%); non-smokers (70%); obese (BMI >30kg/m2; 70%); and did not consume alcohol (90%). Of the purposive sample, 70% had a history of intimate partner abuse: 5% reported current (past year) physical abuse and 45% reported current psychological abuse. The purposive sample did not differ significantly from women in the original study in sociodemographic or abuse characteristics.

Assessment of PTSD symptoms
PTSD symptoms were measured with the Davidson Trauma Scale (DTS), a 17-item, five-point Likert tool with demonstrated reliability and validity. Scores can range from 0 to 136, with an 83 % diagnostic accuracy at scores of 40 or greater (Davidson et al, 1997). The DTS also provides scoring for symptom clusters consistent with the Diagnostic and Statistics Manual for Mental Disorders IV (DSM-IV-R; APA, 1995).

Additional assessment of depressive symptoms was accomplished with the 20-item revised Center for Epidemiologic Studies–Depression (CES-D) scale (Eaton et al, 2003). Scores can range from 0 to 60, with scores of 16 or greater indicative of high depressive symptoms (Carpenter et al, 1998; Eaton et al, 1998). A Cronbach’s alpha of 0.94 in the original study indicated acceptable reliability of the CES-D scale (Woods et al, 2005). Cronbach’s alpha coefficient of reliability is a measure of internal consistency of a psychometric instrument and indicates the extent to which a set of test items measures a single variable. It is generally accepted that 0.70 is an acceptable level for new instruments and 0.80 for established instruments.

Women who scored >40 on the DTS were identified as having PTSD symptoms and women with scores >16 on the CES-D were identified as having depressive symptoms. The instruments measure symptoms and do not form a clinical diagnosis. However, both the CES-D and the DTS have good predictive ability with the DSM-IV-R criteria for diagnoses (Carpenter et al, 1998; Davidson et al, 1997).

Assessment of glycaemic control
A detailed examination of the healthcare provider’s progress notes for each clinic visit over the previous year was conducted as part of the original study. The number of visits ranged from 2 to 30 visits per year (mean: 12.4; SD: 7.7). Medical diagnoses, HbA_1c levels and current medications were included chi-squared and Pearson’s correlations were performed with the computer programme SPSS version 14.0 (SPSS Incorporated, Chicago, US) to examine the relationship of PTSD symptoms with HbA_1c levels.

Results
All women in this sample reported a history of one to five traumas, with an average of 2.5 traumatic events reported to date (median: 2.0; SD: 1.4). Reported traumatic events included partner violence, life-threatening illness (type 2 diabetes), child abuse and witnessing violence as a child.

DTS scores averaged 36.4 (SD: 33.7). Eight women (40%) scored positive for PTSD symptoms. Scores on the CES-D scale ranged from 0 to 50, with a mean of 37.25 (SD: 16.5). Ten (50%) scored positive for depression. Overall, six (30%) had comorbid PTSD and depressive symptoms. Mental health symptoms did not differ significantly from the original sample.

HbA_1c ranged from 5.8% to 16% (mean: 9.3%; SD: 2.45%). Only five women (25%) met the ADA criterion for glycaemic control (HbA_1c <7%). The women presenting with good glycaemic control differed significantly from women without good glycaemic control as follows: mean age was greater (54.4 years versus 47.7 years; P=0.04), mean depression scores were lower (7.2 versus 19.6; P=0.04) and mean PTSD scores were lower (15.2 versus 43.5; P=0.05). Table 2 presents a frequency analysis comparing mental health symptoms by level of glycaemic control.

There was a statistically significant positive association of PTSD symptom scores with HbA_1c levels (r=0.59; P=0.006) and of depressive symptom scores with HbA_1c levels (r=0.560; P=0.01). No other potential confounding variables, including age or BMI, were associated with HbA_1c. Of particular interest, there was a significant association of the avoidance symptom scores (r=0.686; P=0.001) and hyperarousal symptom scores (r=0.478; P=0.033), but not intrusion symptom scores (r=0.270; P=0.25) with HbA_1c (Table 3).

Discussion
Using HbA_1c as a surrogate for medication adherence in people with diabetes, our finding of a statistically significant association of increased reporting of PTSD symptoms with higher HbA_1c levels suggests a biopsychosocial link in understanding diabetes management (avoidance symptom scores: P=0.001; hyperarousal symptom scores: P=0.033;Table 3). These results are similar to other studies that identified PTSD symptoms and medication nonadherence in individuals who have experienced myocardial infarction, transplant surgery, or are HIV appositive (Dew et al 1999; Safren et al, 2003; Shemesh et al, 2004).

In papers published in 2000 and 2004, Shemesh and colleagues theorised that the avoidance symptom cluster (see Table 1) could potentially explain nonadherent behaviour as a mechanism to avoid being reminded of the illness. We also found a statistically significant association between the avoidance symptom cluster and elevated HbA_1c levels. In addition, we found a significant association of hyperarousal symptoms with elevated HbA_1c. There is biological plausibility for this relationship (Lavallo, 2004; Gill et al, 2005):

• Hyperarousal symptoms reflect an excess physiological response.
• This exaggerated response is associated with increases in levels of norepinephrine, thyroid h
• These biophysiological changes, which have been documented in people with PTSD, could reduce the body’s ability to manage hyperglycaemia.
• Biological factors can act in addition to behavioural avoidance of regimen adherence, but the bio-behavioural interactions are likely to be multifactorial and complex.

When reported in addition to PTSD, we found a significant association of depressive symptoms with elevated HbA_1c levels, similar to the findings of Trief and colleagues (2006) who studied male veterans. PTSD is typically comorbid with other psychiatric disorders, the most common of which is depression (Trief et al, 2006). In the study reported here, there was no difference in the glycaemic control status for participants with and without depression-only symptoms (P=NS).

There are some limitations to our findings. This secondary data analysis was comprised of a small sample size and HbA_1c levels were extracted from medical records rather than being drawn concurrently with the mental health evaluation. However, our findings are strengthened by control of other important extraneous factors, such as BMI, age and other medical conditions that could affect glycaemic control. This is the first known study to examine the association of glycaemic control and PTSD symptoms among women in a primary care setting. Generalisability of the findings is limited by the homogeneity of sample. Participants were primarily African–American, from urban areas, of low-income and uninsured. This study should be repeated in other samples.

Conclusion
Nurses who manage people with diabetes should consider incorporating a holistic biopsychosocial perspective in the plan of care. This research supports the concept of including a mental health evaluation, particularly for PTSD and depression. Nurses should screen for intimate partner violence, as this is a psychosocial risk for PTSD and depression (Campbell, 2002). There are no universally accepted methods of screening for PTSD, depression or other psychiatric comorbidities. However, a four-question screening tool has been developed for women who are victims of partner violence, with good predictive values for depressive symptoms (96%), PTSD symptoms (84%) and suicidal ideation (54%; Houry et al, 2007). People with mental health symptoms should be referred to their primary care provider or a psychiatrist for diagnosis and treatment.

Specific to psychosocial stress in people with diabetes, the Diabetes Distress Scale consists of 17 items that assess emotional burden, regimen-related distress, physician-related distress and diabetes-related interpersonal distress (Polonsky et al, 2005). Further research is needed to test the effects of screening and intervention on degree of glycaemic control.

Further research is also needed to better understand the biopsychological mechanisms that contribute to glycaemic control, as well as randomised controlled trials to evaluate the incorporation of mental health treatment into a holistic diabetic management plan.

Acknowledgements
This study was supported by NRSA 1 F31 NR07600-01, Sigma Theta Tau-NuBeta Chapter dissertation grant and the Independence Foundation.

Editor’s commentary
This article is unique and interesting; however, I believe that it only confirms what healthcare professionals in the UK have believed for years – that is that having diabetes, regardless of type, is a difficult condition to self-manage even when you are feeling at peace with the world.

The authors mention that less than half the people with diabetes in the US achieve target HbA_1c levels. When there are other factors that affect behaviour and self-care strategies such as depression or low self esteem, unfortunately, self care seems to suffer. When there are additional stressors such as physical or mental abuse, as mentioned in the article, then a lack of self care manifested by a high HbA_1c is not surprising. A person who is in fear of daily physical abuse is not focused on managing their diabetes: I believe this is because they do not worry about long-term complications but focus on the prevention or avoidance of the next act of abuse.

In the UK, as part of the Quality Outcomes Framework introduced within the General Medical Services GMS contract in 2004 and updated in 2006, there are now points attached to depression identification in diabetes that should be included in the annual review process. This is highlighting the fact that depression is a highly under-reported symptom of living with a chronic disease. At least now there is an official prompt to explore the presence of depression, which may uncover the poor reporting of depression among people with diabetes.

As these questions are attached to points and worth money, there is an incentive to ensure that they are asked by the healthcare professional. Unfortunately, there is still a stigma attached to mental health problems and people find it difficult to admit that they are not coping with, or are failing to manage, their diabetes. It is up to all the healthcare professionals working in diabetes care to be sensitive to small signs and symptoms that may indicate depression, and create the right atmosphere in which an individual feels able to divulge coping difficulties.
Debbie Hicks, Nurse Consultant – Diabetes, Enfield PCT