The present study examined a quantitative assessment of symptoms that were hypothesized to be associated with the decision to initiate or defer HCV treatment initiation, independent of standard of care clinical considerations. The assessment was part of a multidisciplinary approach implemented within a comprehensive HIV primary care clinic caring for more than 3,000 patients of which 27% are co-infected with HCV.
We used a self-administered computer assisted symptom questionnaire completed just prior to the patient-provider encounter. Patient reported symptoms are important components of comprehensive clinical assessment [22
]. Recently, a study of cancer patients found that internet assessment of different symptoms supplement traditional office visit discussions and fill important gaps in clinicians' knowledge, significantly improving patient safety and quality of care [23
]. This suggests improvement in sensitivity of ascertainment when traditional methods are replaced by technologically enhanced symptoms assessment [24
The study results suggest that the HCV symptom inventory could be a useful tool for consideration of HCV treatment eligibility in HIV infected patients. The HCV symptom inventory has a meaningful internal structure with excellent reliability as well as initial evidence supporting construct, convergent and predictive validity. The underlying structure of the inventory was factor analyzed into 3 primary constructs reflecting neuropsychiatric, somatic and sleep symptoms.
Similar to other reports [5
], the present study found that severe neuropsychiatric symptoms and high CES-D scores at baseline were associated with independent judgment regarding treatment candidacy, perhaps due to clinician awareness of the likely worsening effect that interferon may have on neuropsychiatric symptoms [27
]. A novel observation was that patients with worse somatic and sleep symptoms at baseline were less likely to be selected for HCV therapy initiation. Our results highlight the importance of baseline screening for somatic and sleep symptoms, particularly those that are known to be aggravated by HCV therapy: body aches, different somatic non-specific pains, skin complaints and nausea. Our results are in agreement with prior results validating the CES-D scale for HCV therapy and also identifying a somatic component in its underlying structure [21
]. However, the CES-D somatic component has only 3 items (poor appetite, keep in mind what I was doing, and everything an effort to do) and does not focus specifically on comprehensive symptom evaluation.
The HCV symptom inventory could be viewed as a screening tool complementary to a well validated depression scale such as CES-D during HCV treatment eligibility assessment. Recognizing, acknowledging and medically assessing the presence of somatic and sleep symptoms in addition to neuropsychiatric screening prior to HCV therapy in HIV patients may enhance patient engagement in care and compliance with HCV therapy.
We found no effect of HCV genotype, viral load (both of which are predictors of HCV treatment response), CD4 cell count, HIV risk factor and gender on HCV treatment candidacy in the present study. This is consistent with the inclusive approach of the study team to treat as many HCV/HIV co-infected patients as possible. Our program aim is to also treat homeless individuals, patients with past and current substance use, and those with ongoing psychiatric conditions as long as they remain engaged in care. Consistent with prior reports [5
], in our cohort whites tend to be treated more frequently that non-whites in unadjusted analysis. However, after adjusting for HIV viral load and neuropsychiatric symptom score, race was no longer associated with a treatment initiation decision. This likely reflected the clinician's perception of the decreased likelihood of HCV treatment response in non-white individuals when other known factors are taken in consideration [29
]. The study was conducted before the interleukin-28B gene polymorphism was widely available as a screening tool to further characterize likelihood of HCV treatment response among non-white patients [30
]. We noted that patients with higher HIV viral loads were less likely to be treated, but these patients had recently re-initiated HIV therapy. Interestingly, we did not find a significant negative impact following HCV therapy initiation in sleep symptoms or CES-D scores; these treatment-emergent symptoms were routinely treated by either the treatment team or the primary care providers, thus attenuating the expected effects on self reported depressive and sleep symptoms.
Inference from the study is subject to several limitations. First, since only 74% of patients completed the symptom inventory at each HCV staging visit, bias by unmeasured patients with worsening symptoms cannot be excluded. However, there was no difference in the mean symptom scores between patients who completed the symptom inventory at every clinic visit and those who completed the inventory occasionally (48.3 vs. 45.1, p = 0.79). We believe the missing inventories were more prevalent at the beginning of the implementation of our new clinic model when patients were not familiar with the process. We have subsequently implemented a procedure whereby the medical assistant immediately notifies our substance abuse counselor when a patient is placed in the exam room to avoid delays in the completion of the symptom inventory. Second, we recognize that both measured and unmeasured co-morbidities may increase variability in symptom scores. However, our immediate study objective was to examine the extent to which systematically measured symptomatology predicts treatment initiation decisions, while our longer term objective is to assess the symptom inventory as a predictor of adherence and treatment completion. Third, although the sample size could be considered relatively small (n = 126) it was sufficient to demonstrate the reliability and preliminary validity of the symptom inventory. Fourth, although stepwise multiple regression methods are popular in medical research, we acknowledge that they may have potentially serious shortcomings for valid inference [32
]. As we described above this is an exploratory analysis to validate the symptom inventory and not a randomized controlled clinical trial. Future longitudinal trials are needed to confirm of our model conclusions. Finally, because the symptom inventory was performed only in English, inference regarding the observed findings is limited only to our English-speaking population. However, we recently have implemented a Spanish version of the symptom inventory for our patients when needed.
In conclusion, a 41 item hepatitis-related symptom inventory was found to have a clinically meaningful 3-factor structure with excellent internal consistency reliability and predictive validity. Both symptom and CES-D scores were predictive of treatment initiation decisions. Future research will identify whether the symptom inventory is useful in predicting adherence and treatment completion.