Amyotrophic lateral sclerosis (ALS) is a progressive neurodegenerative disease characterized by the death of motor neurons. ALS results in difficulty swallowing, paralysis, and respiratory dysfunction [1
]. Riluzole and edaravone are the two primary medications prescribed to slow the underlying ALS etiology and are the only two currently approved by the United States Food & Drug Administration. At best, both medications slow disease progression for only a few months [2
Since ALS is a debilitating and fatal disease, quality of life (QoL), an individual’s sense of well-being and ability to carry out activities of daily living, is a major concern and driver of clinical care [3
]. Due to the limited efficacy of ALS etiological drugs, palliative interventions are commonly used to alleviate non-curable ALS symptoms with the purpose of improving QoL and/or extending survival duration. Interventions largely consist of modes to manage symptoms such as gastrostomy tubes to provide nutritional supplementation in patients with dysphagia; noninvasive ventilation (NIV) such as bi-level positive airway pressure (Bi-PAP) to improve respiratory efficiency [5
]; equipment to assist in ambulation such as leg braces, walkers, or wheelchairs; oral nutritional supplements like vitamins, canned meal supplements, or alterative herbs; and palliative pharmacologic treatments, such as antidepressants, antianxiety, antispastics, anticholinergics, and pain medications [6
]. Theoretically, these therapeutic interventions aid patients in living a more comfortable and fulfilling life, which is hypothesized to improve patient mood. Studies have demonstrated that poorer mood is correlated with faster disease progression and shorter survival duration [7
]. Thus, palliative therapeutic interventions are hypothesized to indirectly increase survival duration [7
Not surprisingly, both functional metrics and QoL metrics are useful in assessing the overall impact of ALS, both physically and psychologically, in the context of patient care and clinical trials [11
]. Several generic QoL instruments are utilized to assess patients with ALS, such as the McGill Quality of Life Scale (MQOL), Short Form 36 (SF-36), the Neurology Quality of Life (Neuro-QoL) Measurement System, the Sickness Impact Profile, the Schedule for the Evaluation of Individual Quality of Life (SEIQoL), among many others [11
]. There are a few ALS disease-specific instruments used as well. The revised ALS Functional Rating Scale (ALSFRS-R) provides an overall measure of functional disease progression [17
], which encompasses respiratory function and activities of daily living. In contrast, the ALS Assessment Questionnaire-40 (ALSAQ-40) and the ALS-Specific Quality of Life Questionnaire (ALSSQOL) [11
] focus on quality of life.
The existence of numerous instruments reflects the complexity of QoL and how difficult it is to measure due to its subjective and multifactorial nature, as well as the diverse definitions for QoL [19
]. QoL assessment is further complicated in ALS due to underemployment of QoL assessments, which are often lengthy, take much effort to compete, and can be skewed by patient or caretaker bias [20
]. There is also a lack of standardized QoL instruments in the clinical setting and, some believe, overemphasis of strength and physical factors among current instruments [18
]. Rather than using a survey, this study utilizes a newer metric, clinical impression of mood (CIM) [20
]. CIM relies on clinician evaluation of patient mood during each visit using objective verbal and non-verbal cues, as well as subtle subjective cues based on the clinician’s consistent, long-term relationship with the individual patient over three or more visits. CIM uses text mining of the electronic medical record to map specific adjectives to categorize the mood as either “positive” or “negative” [20
]. Past research lends support to the feasibility of such a metric, as depression, for example, has several facial and vocal biomarkers that can be easily assessed [26
An important component in QoL is the patient’s mood throughout the course of the disease. There has been some evidence that higher QoL improves survival in patients with terminal prognoses [28
]. There is much interest in the impact of mood-stabilizing drugs on patient QoL and survival. However, there is evidence that clinical depression among ALS patients is not as widespread as might otherwise be expected [29
], and that antidepressants do not always significantly change ALS patients’ moods [30
]. In fact, impact of antidepressant is controversial based on preclinical mouse model experimental results [31
]. For example, some antidepressants have been shown to delay disease onset and extend the lifespan of mouse models with ALS and other neurodegenerative diseases [32
]. Selective serotonin reuptake inhibitors (SSRIs) have specifically shown preclinical pro-cognitive effects [34
]. However, antidepressant usage is not ubiquitously positive. The (SSRIs) sertraline and paroxetine reduced astrocyte viability, induced dose-dependent intracellular calcium elevation, and activated apoptosis in primary astrocyte and neuron co-cultures [35
]. In a clinical setting, antidepressant users were more likely to develop dementia than non-users, regardless of their depressive state [36
]. There is not enough conclusive scientific evidence to determine whether the potential QoL or neuroprotective effects of antidepressants are strong enough to improve survival in ALS patients.
Beyond antidepressants, other prescription or over the counter palliative drugs are used to treat some symptoms of ALS, such as excessive secretions or drooling (e.g., with anticholinergics), muscle spasticity or fasciculation (e.g., with antispastics like baclofen), supplements for nutritional support or to combat ALS-related oxidative stress (e.g., multivitamins or specific vitamins like vitamin E), and pain medications (e.g., ibuprofen, acetaminophen, and limited use of some opioids). ALS patients generally have less antecedent conditions (defined as conditions present prior to ALS onset) compared to age-matched controls [37
]; nonetheless, a substantial portion of ALS patients are treated for antecedent conditions that are common in the age-matched general population, such as hyperlipidemia (e.g., with statins), hypothyroidism (e.g., with levothyroxine), hypertension (e.g., with calcium channel or beta blockers), and diabetes (e.g., with insulin replacements or blood sugar regulators like metformin). Treatment of antecedent conditions has also been controversial in the literature, especially treatment of high cholesterol [39
] or diabetes, where there is some evidence these treatments could expedite ALS etiology. Likewise, use of pain medications (namely opioids) [41
] and some categories of antidepressants, which have side effects that induce or exacerbate respiratory depression, are also controversial in ALS patients. Most ALS patients already have compromised respiration due to the paralyzing effects on ALS on the diaphragm and intercostal muscles used for breathing. Thus, even small increases in respiratory depression from palliative treatment could be perceived as a risk [43
Weighing the potential benefits of improved QoL against the risks of potentially decreasing ALS survival duration is necessary. More analysis is needed to better understand the impact of ALS patient mood and its relationship to physical function and to better understand the quantifiable impact of palliative interventions on ALS survival duration. Discordance among existing studies warrants additional comprehensive evaluations to assess existing hypotheses and to generate new ones. The aims of this study are to: (1) Discern the relationships between physical health, mood, and survival duration in ALS; (2) evaluate the usage of common QoL medications like antidepressants and pain medication on ALS patient mood and survival duration; (3) assess prevalent pharmaceutical and supplement usage to determine potential beneficial or harmful associations with ALS survival duration. A better understanding of these quality of life and survival duration associations can improve ALS clinical care.
2. Materials and Methods
This is a secondary analysis of a retrospective cohort study comprising 8028 clinical visit records collected from 1585 patients at the Emory ALS Clinic (Emory University Hospital, Atlanta, GA, United States) into a FileMaker Pro relational database. Data collection, organization, and quality control methods are as previously published in prior work with this data set [5
]. All data was de-identified and anonymized. The study was conducted in accordance with the Declaration of Helsinki; the protocol was jointly approved by the Institutional Review Boards of Emory University and Georgia Institute of Technology (Protocol H16257).
2.1. Temporal Assessments and Patient Characteristics
Temporal assessments are used to assess associative relationships. ALS start point and end point is used to calculate the most important outcome metric for any ALS assessment, which is survival duration. Survival duration is defined as time passed between ALS start point and end point. ALS end point is defined as the date of death, which is a straightforward determination. ALS start point is defined as the date of initial ALS diagnosis by a physician.
Patient characteristics assessed as part of this study include gender, ALS onset type, and ALS onset age. Gender is defined as biological gender (male or female). ALS onset type is defined as limb, bulbar, or other/unclassifiable [45
]. Table 1
illustrates the patient characteristics of the overall cohort. The majority of the patients resided in the southeast United States.
For the pharmaceutical and supplement portion of this study’s analysis, included patients were further stratified based on their ALS onset age relative to the literature mean of 55 years (e.g., early onset is <55 years and late onset is ≥55 years) and ALS survival duration relative to the pharmaceutical & supplement ALS cohort mean of 3.67 years (e.g., short survival duration is <3.67 years and long survival duration is ≥3.67 years).
2.2. Assessment of Clincal Impression of Mood
Initial inclusion criteria for the clinical impression of analysis (CIM) analysis consisted of patients who had a specialist-determined ALS diagnosis and ≥5 clinic visits to see the same ALS clinic specialty team. For each clinic visit, the clinician denoted a short, qualitative description of the patient’s exuded mood, referred to as clinical impression of mood (CIM), as previously published [20
]. Text mining of the study database was used to classify clinical impression of mood (CIM) using a binomial score of 0 or 1. A perceived positive or neutral mood for the visit day was assigned a “0”, whereas a perceived negative mood was given a value of “1”. Note that CIM scores for patients with documented pseudobulbar affect (PBA) were excluded because visualized emotion and facial expression of a PBA patients, by clinical definition, does match the patient’s internalized mood during a PBA event.
Text mining automation with a synonym table (Table S1
) was utilized to initially and agnostically classify mood according to the previously published method for CIM [20
]. However, more ambiguous adjectives (shown in rightmost column of Table S1
), required more context to make a determination; for these, three trained experts reviewed the full record before assigning a CIM score for the visit. Individual clinic appointments with missing data or incomplete data on patient mood were excluded from the CIM analysis.
CIM was utilized in the present study due to a lack of standardized quality of life (QoL) surveys for this data set. The advantage of CIM is the lack of patient or caretaker survey bias [20
]. The limitation of CIM is the reliance on clinician observed interpretation of verbalized communication or non-verbalized facial expression as an assessment of patient mood.
2.3. Assessments of Physical Health and Modulators of Quality of Life
Forty assessments of physical health or modulators of quality of life were classified into the following categories: Respiratory, pain, disability, muscle control, oral muscle control, vocal control, PEG, therapy, QoL medication, depression, sleeping problems, and social. Thirty-four of the assessed factors are of the binary data type, meaning they relate to either a yes or no assessment. For example, a condition is either “present” or “not present”, or a therapy either “used” or “not used” by the patient. Six of the factors are of the continuous data type, meaning they have a quantitative value assigned from a continuous scale. The continuous factors include standard quantitative measures of respiratory function, as well as the revised ALS functional rating scale (ALSFRS-R) [17
]. The ALSFRS-R includes two main components, a survey assessment of activities of daily and an assessment of respiratory function; a higher ALSRFRS-R total score or sub-score indicates more function, whereas a lower score correlates with more severe disease progression [17
]. The present study uses the ALSFRS-R total score and the ALSFRS-R respiratory sub-score. Most of the assessed physical health factors are standardly measured in ALS, well-defined, and often used by the field in ALS predictive analysis [45
The categorizations in Table 2
make most of the assessed factors self-explanatory, but a few warrant further explanation. PEG, an acronym for percutaneous endoscopic gastrostomy, is more commonly known as a “feeding tube”; this surgical intervention assists in dysphagia and corresponding weight loss by providing a vehicle for regular nutrition and hydration directly to the stomach. Cough assist and suction are two different interventions to assist with secretion clearance, which is impaired in ALS due to respiratory muscle paralysis. Bi-PAP is an acronym for bi-level positive airway pressure, a form of non-invasive ventilation to improve respiration. Assistive devices include devices that improve activities of daily living such as a wheelchair, walker, leg brace, cane, etc.
Medications that are known to be modulators of quality of life were included in the CIM analysis as binary factors (“use” or “did not use”). As shown in Table 2
, these QoL modulators include antidepressant medication usage, drooling medication usage (namely anticholinergics), non-opioid pain medication usage (prescription medication for pain that does not target opioid receptors), opioid pain medication usage (prescription pain medications that target opioid receptors), NSAID medication usage (common over the counter pain medication, like ibuprofen, used to combat pain and inflammation), sleeping medication usage (prescription or over the counter drugs or supplements to induce sleep or treat clinical insomnia), and muscle control medication usage (prescription muscle relaxants or antispastics, like baclofen). Usage of these medications were assessed to determine their potential relationship with CIM. Antidepressant usage was further broken down into sub-categories based on mechanism of action: SNRI, SSRI, tricyclic antidepressant, and general antidepressant.
2.4. Other Pharmaceutical and Supplement Usage
A separate analysis was performed to assess all pharmaceutical and supplement medication usage and their associative relationships to ALS onset age, gender, and survival duration. All pharmaceutical or supplements were either prescribed by the ALS clinic or were reported as being used by the patient at their clinic visit (e.g., could be drugs prescribed from another healthcare provider or the patient’s own purchase of over the counter drugs or nutritional supplements). The medications included palliative medications for ALS symptoms, treatments for other antecedent or co-morbid health conditions, and vitamins or nutritional supplements for improved general health or prophylaxis. Medication usage in the present study was evaluated on a per-patient basis. Medications were categorized by pharmacological class, active ingredient, or intended purpose. Based on statistical power for usage sample size (e.g., number of patients using the medication), the top 8.5% of medications were included in the analysis, which equates to 47 medication categories (Table S2
2.5. Data Analysis and Statistical Methods
Four analyses were performed: (1) Analysis to assess CIM and its association to physical health factors; (2) analysis to the association of ALS survival duration with CIM, physical health factors, and known modulators of QoL; (3) analysis to assess the association of CIM with antidepressant usage, including overall antidepressant usage versus non-usage, usage of specific antidepressant sub-classes based on mechanism of action, and number of distinct antidepressants jointly co-used; and 4) analysis to assess the association of the most commonly used pharmaceuticals and supplements in the study cohort with ALS survival duration.
For analyses 1–3 outlined above, a Chi-square test was used to assess correlation between the factors with a binary data type, and a generalized linear model (GLM) was used to assess correlation between CIM and factors with a continuous data type. The GLM type was a binary logistical regression. A standard t-test was used to determine if antidepressant usage or non-usage impacted survival duration. A one-way ANOVA analysis was used to determine if a specific antidepressant class had a significant effect on a patient’s CIM.
For analysis 4, two methods were utilized to assess associations between the usage of a specific medication or supplement category and patient gender, ALS onset age, and survival duration. The Fisher’s exact test was used to determine statistical significance of association of each medication class’s usage with survival duration. Conditional random forests modeling was utilized to assess the importance of pharmaceutical medications and supplements on survival duration. Conditional random forests were chosen over standard random forests to address potential class imbalances in the data [46
]. The conditional random forests model was simulated four times and the variable importance was averaged to determine the final result. The area under the curve (AUC) of variable importance was used as the performance metric [45
Confidence intervals of 95% and 99%, respectively, were used for comparison of significance. Associations throughout the results are designated as insignificant (p-value > 0.05), low threshold of significance (0.05 > p-value ≥ 0.001), or high threshold of significance (p-value < 0.001). Recall that significance threshold does not imply differences in the magnitude or strength of the association. Rather, the threshold of significance only represents the probability of a false positive result. All statistical analysis was performed in Matlab™ (the MathWorks, Inc., Natick, MA, USA).
The primary goal of this study was to assess associations among patient mood, quality of life modulators, usage of other pharmaceuticals and supplements, and ALS patient survival duration. Mood was assessed using the clinical impression of mood (CIM) metric for each clinic visit [20
]. Quality of life modulators included palliative interventions for ALS symptoms; QoL medication usage for depression, pain, and sleep; and factors related to patient social support. Assessment of other pharmaceuticals and supplements examined usage of the most prevalent treatment modalities, regardless of indication, whether for ALS-related symptoms, antecedent disease, co-morbid disease, or general health prophylaxis. Thresholds of statistical significance are denoted for p
< 0.05 and p
< 0.001 throughout the results.
3.1. Factors Assessed for Association with Clinical Impression of Mood (CIM)
lists the 40 health factors analyzed for their correlation with CIM. Eleven of these factors have significant associations with CIM as illustrated in Table 3
where the * indicates a significance threshold where 0.05 > p
-value ≥ 0.001, and ** denotes a significance threshold where p
-value < 0.001. The significant factors associated with CIM included both assessment of physical factors as well as QoL modulators.
3.2. Physical Health Factors Significantly Associated with Clinical Impression of Mood (CIM)
Three physical health factors met the high threshold for significance (p
< 0.001): FVC, % predict FVC, and jaw jerk. FVC and % predict FVC both measure respiratory output. Higher scores correlate to better respiratory function. Thus, unsurprisingly, higher FVC and FVC % predict result in lower CIM (e.g., moods that are happy are neutral). The presence of jaw jerk results in higher CIM (e.g., more depression or anxiety). Other physical health factors that had an association meeting the lower threshold for significance (0.05 > p
-value ≥ 0.001 included: Toe walk, dysphagia, drooling, ALSFRS-R total score. Ability to do the toe walk represents more physical function and is associated with lower mean CIM. Presence of dysphagia or drooling resulted in a higher mean CIM. A higher ALSFRS-R total score represents a combination of better respiratory function and/or better muscle function to independently perform activities of daily living [17
]. A higher ALSFRS-R total score is associated with a more positive mood, equating to a lower mean CIM.
3.3. Quality of Life Factors (QoL) Significantly Associated with Clinical Impression of Mood (CIM)
Three quality of life (QoL) factors had associations that met the high threshold for significance (p < 0.001): Presence of general pain and the use of therapeutic devices, namely Bi-PAP and cough assist. One QoL factor, no family reported, had an association that met the low threshold for significance (0.05 > p-value ≥ 0.001). Not surprisingly, presence of pain is associated in a more negative mood (depression or anxiety), equating to higher mean CIM. Interestingly, use of the Bi-PAP, cough assist intervention was actually associated with a positive mood (happy or neutral), equating to a lower mean CIM.
3.4. Physical Health Factors Have Strong Association with Survival Duration
The physical health factors with associations to shorter survival duration that met the high threshold of significance (p < 0.001) are: Presence of disability, presence of fasciculation, presence of jaw jerk, presence of dysarthria, lower ALSFRS-R scores (corresponding to less independent function), forced vital capacity (FVC), percent predicted forced vital capacity (FVC % predict), lower ALSFRS-R respiratory sub-score (corresponding to less respiratory function), and negative inspiratory force (NIF). Physical health factors with association to shorter survival duration meeting the low threshold of significance (0.05 > p-value ≥ 0.001) are presence of dysphagia, presence of tongue atrophy, and presence of head drop. These results are consistent with other prior studies.
3.5. Quality of Life (QoL) Modulators Have Strong Association with Survival Duration
Several quality of life (QoL) modulators had associations with survival duration that met the high threshold for significance (p < 0.001): CIM, general pain, cough assist usage, Bi-PAP usage, PEG (“feeding tube”) usage, and not being accompanied to appointment. More specifically, lower CIM (e.g., patients with a more positive or neutral mood), lack of general pain, Bi-PAP usage, PEG usage, cough assist usage, and patients being accompanied to their appointment, are all factors that are individually associated with longer ALS survival duration.
Other QoL factors with associations to longer survival duration that met the lower threshold of significance (0.05 > p-value ≥ 0.001) are: Suction usage, lack of reported sleeping problems, reported family support, reported friend support, antidepressant medication usage, sleeping medication usage, muscle related medication usage, and non-opioid pain medication usage.
Overall, usage of most QoL medications and interventions by ALS patients as well as the general presence of social support was associated with a significant increase in survival duration. The QoL factors in Table 2
that had an insignificant association with survival duration (p
> 0.05) are drooling medication usage, opioid medication usage, and NSAID medication usage.
3.6. Antidepressant Usage Does not Impact Clincial Impression of Mood (CIM)
Approximately 34% of the patients meeting inclusion criteria for the CIM analysis took one or more antidepressants. Given that antidepressants are prescribed to improve depression or anxiety, it could be hypothesized that antidepressants may have an impact on the clinician-visualized CIM at the clinic visits. To assess this hypothesis, the association between mean CIM score and antidepressant usage was compared.
“Antidepressant usage” was defined as regular usage of antidepressants over three or more visits where CIM was also measurable. Figure 1
illustrates there is an insignificant change in CIM (p
> 0.05, t
-test) with the usage of antidepressants. To assess if antidepressant impact on mean CIM score could be class-specific, antidepressants were divided into their respective pharmacological classes based on mechanisms of action: SSRI, tricyclic, and SNRI. Still, there was no significant association with any class of antidepressant and mean CIM score (Figure 2
> 0.05, one-way ANOVA). Finally, the number of different antidepressants being used by each patient was analyzed. The usage of multiple antidepressant medications also showed no association (p
> 0.05) with mean CIM (Figure S1
It should be noted that the deviations shown on Figure 1
and Figure 2
(and Figure S1
) for completeness are visually deceiving due to the binary nature of the CIM metric. The statistical variance for all included metrics was actually overall much lower for this large ALS cohort study compared to smaller ALS cohort studies. Nonetheless, there is more variance in mood compared to other physical health metrics. The increased variance in mood is because mood can vary from clinic visit to clinic visit, whereas physical progression in ALS essentially continuously declines with each subsequent clinic visit.
3.7. Visualisation of Medication and Supplement Usage Associations
visualizes the covariance and the usage prevalence of each intervention category. Moreover, it specifically shows the connections between intervention usage and ALS patient characteristics (gender, onset age, longer survival duration, shorter survival duration). The shorthand nomenclature for Figure 3
is defined in Table S2
. The connections demonstrate the degree that each patient characteristic connects with each intervention. Patients with a longer survival duration have more connections to interventions than those with shorter survival duration. Note that “long” or “short” survival duration is relative to the average survival duration, 3.67 years, for the ALS patients included in the pharmaceutical and supplement cohort analysis. Interestingly, males and females have roughly the same number of intervention connections despite males outnumbering females at an approximate 3:2 ratio. Thus, female patients use more interventions per capita. While a few medications are female-specific, the presence of female-specific interventions (such as progesterone) do not account for the overall higher usage prevalence of multiple simultaneous interventions by females in the patient cohort.
3.8. Statistical Assessment of Medication and Supplement Associations
Patient characteristics and interventions characterized as medications or supplements were used to predict survival duration using both Fisher’s exact test and conditional random forests. For these analyses, non-medication interventions like Bi-PAP usage, cough assist, and social support were not included.
First, Fisher’s exact test was used to determine the significance of the association of medication usage with survival duration. Of the 47 medication and supplement categories (see Table S2
for detailed definitions), 17 were associated with a significantly longer increase in survival duration (p
< 0.05) ranging from +4.5 months to +16.5 months. No oral medication or supplement was found to result in a significant decrease in survival duration. Three medications had associations with longer survival duration meeting the high level significance threshold (p
< 0.001): Alternative herbal medicines (+9.6 months), Vitamin A (+8.9 months), and multivitamins (+8.2 months). Table 4
illustrates the 17 medication or supplement categories with significant associations (p
< 0.05) with longer survival duration, and Table S3
illustrates the 30 medication or supplement categories, which had insignificant associations with survival duration.
Of the 17 interventions associated with longer survival duration, only five interventions were found to have a significantly higher usage prevalence (p < 0.05) on the basis of onset age; all five interventions (Vitamin A, muscle relaxants, antispastics, [non-specific] vitamin, and anticholinergics) were more prevalently utilized by the younger onset age group (onset age < 55 years). Seven of the 17 interventions that statistically increased survival duration had a statistically higher usage prevalence (p < 0.05) on the basis of gender; all nine categories (stimulant, stimulant laxative, nutritional supplement, antihistamine, [non-specific] vitamin, non-opioids, anticholinergics, and muscle relaxants) were more prevalently utilized by females than males. Finally, four of the 17 interventions that statistically increased survival duration had a statistically higher usage prevalence (p < 0.05) on the basis of onset type. Four interventions were more prevalently used by limb onset (muscle relaxants, antispastics, vitamin A, [non-specific] vitamin), whereas one intervention was more prevalently used by bulbar onset (anticholinergics).
The Fisher’s exact test alone, does not account for complex interactions that influence survival duration. However, random forests do include complex interactions between all variables when determining which are most predictive of the response (e.g., survival duration). Note that, while other non-drug interventions were excluded in the random forests analysis, the feeding tube was included in the random forests because it corresponded to supplement usage. Nonetheless, feeding tube supplement usage was separated from the nutritional supplement category, the latter which encompassed prepared meals or canned nutrition drinks.
illustrates the relative area under the curve (AUC) performance for variables that best predict survival duration using conditional random forests modeling. The “best” predictive variables included three patient characteristics (ALS onset age, ALS onset type, and gender) and eight medication & supplement categories. The collinear onset age and onset type are, unsurprisingly, the top two predictors whereas gender, the least predictive of the patient characteristics, was ranked 11th. The eight in-between predictors of disease duration, respectively ranking 3 through 10, are all medication or supplement categories: Vitamin A, multivitamin, feeding tube supplements, alternative herbal medicine, antihistamine, muscle relaxant, stimulant laxative, and antispastics. All eight medications & supplements have a higher usage prevalence in the longer disease duration population. Notably, the eight medications & supplements predicted as most important from the conditional random forests were also predicted to be significant at the high threshold (p
< 0.001) in the Fisher’s exact test.
The results of this large, retrospective cohort study illustrate the significant associations of patient mood, modulators of quality life, and multiple non-ALS specific pharmaceuticals and supplements on ALS patient survival duration. While there is not yet a cure for ALS, this study illustrates multiple practices that can improve ALS patient quality of life and corresponding survival duration.
The results of this study underscore the importance of patient mood and its significant relationship to survival duration. Not surprisingly, measures of ALS progression do significantly contribute to worse patient mood. However, presence of pain and lack of social support also significantly contribute to worse patient mood. Being accompanied to a clinic appointment and the presence of family or friend support were found to be critical for significantly improving patient mood; this type of social support could be increased through ALS social programs and/or better awareness among families and caretakers on the importance of social support in patient outcome. Interestingly, the use of antidepressants appears to have no effect on ALS patient mood in this large cohort study. This could be due to changes in the etiology underlying depression in the ALS neuropathology, or it could be that the sources of poor ALS patient mood are not readily treatable with antidepressants.
The results of this study identify several therapies, pharmaceuticals, and supplements with significantly positive association with survival duration, which warrant changes in clinical practice. The novel finding that significant survival-enhancing respiratory interventions (Bi-PAP and cough assist) are also associated with a significantly overall better patient mood should greatly increase the prevalence of physician prescription and patient usage of these therapies. Additionally, the negative association of pain with ALS patient mood and the large, positive association of pain medication usage with survival duration, especially non-opioids, indicates pain management should be more at the forefront of ALS patient care. Fasciculation and muscle spasms are typically considered only palliative symptoms, but could have a more complex role in ALS outcomes; both antispastics and muscle relaxant usage was significantly correlated with longer survival in all analyses, whereas the presence of untreated fasciculation and spasms correlated with shorter survival. Finally, the results of this study illustrates that there should be more emphasis on ALS patient health prophylaxis with nutritional supplements, vitamins, and alternative herbs, as they strongly correlated with longer survival duration. Feeding tube usage (PEG) also increased survival duration and did not negatively impact patient mood in the present cohort.
There were a few interesting results that do not yet implore clinical recommendation but do implore more investigation. Specifically, it is interesting that antihistamines and tricyclic antidepressants significantly increased survival duration, ranking among the highest in all analyses, without a clear connection to ALS symptoms or etiology. Additionally, the potential contraindication of diuretics, antihypertensives, diabetic drugs, and hyperlipidemia drugs is supported by insignificant but qualitative associations with decreased survival duration. The complex interactions of antecedent disease and the treatment of antecedent disease with ALS onset and progression warrants further study. Finally, the results of this study illustrate that additional investigation of “off label” drugs for synergistic benefits in ALS could further improve ALS patient quality of life while we await a cure.