Predictors and clinical implications of shivering during therapeutic normothermia

Abstract

Background

Shivering during induced normothermia (IN) remains a therapeutic limitation. We investigated potential risk factors and clinical implications of shivering during IN.

Methods

Post hoc analysis was performed on 24 patients enrolled in a clinical trial of an automated surface cooling system to achieve IN. Hyponatremia was defined as serum levels <136 mmol/L and hypomagnesaemia as levels <1.5 mg/dL. Continuous heat energy transfer (kcal/h) was averaged hourly. Glasgow Coma Scale (GCS) scores were recorded every 2 h. Shivering status was documented hourly. Mixed effects modeling was used to determine clinical measures associated with shivering. Generalized estimating equation (GEE) models were used to compare baseline-adjusted repeated-measures GCS scores.

Results

About of 24 (39%) patients demonstrated shivering. Shivering was associated with men (67% vs. 21%, P = 0.03), hyponatremia (44% vs. 7%, P = 0.03), and hypomagnesaemia (56% vs. 7%, P = 0.02). The average kcal/h (158 ± 645 kcal/h vs. 493 ± 645 kcal/h, P = 0.03) was greater in shivering patients. Shivering was positively associated with increases in heart rate (P < 0.001), respiratory rate (P < 0.001), and kcal/h (P < 0.001). Non-shivering patients showed a greater increase from baseline GCS (GEE, P = 0.02) at 24 h. No differences in sedative doses or fever burden were noted between shiverers and non-shiverers.

Conclusions

Men, hyponatremia, and hypomagnesaemia may predispose febrile patients treated with IN to shivering. Shivering dramatically increases the amount of heat transfer required to maintain normothermia, and may be associated with adverse effects on level of consciousness.

Appendix : A Calculation of Heat Energy Transfer

In equation form:

$$ {\hbox{q}}_{\rm{u}} {\hbox{ = 60 }} \times {\hbox{ }}\sigma{\hbox{ }} \times {\hbox{ Q }} \times {\hbox{ Cp }} \times {\hbox{(T}}_{\rm{r}}-{\hbox{T}}_{\rm{s}} {\hbox{) }}$$

where q_u = uncorrected heat transfer rate (kcal/hour), σ = density of water (=1 kg/l), C_p = heat capacity of water (= 1 kcal/(kg^oC)) and 60 = conversion factor (per minute to per hour). As defined in the above equation heat being removed from the patient is positive.

The heat transfer rate then was corrected for ambient losses. The ambient heat losses were comprised of energy lost from the fluid supply lines and energy lost from the Energy Transfer pads. To measure the ambient heat loss from the fluid supply line, calibrated in line thermistor probes were attached to the supply and return ports of the Arctic Sun and the proximal end of fluid supply line was attached to these probes. Shunts were connected to three of the five sets of connector ports at the distal end of the fluid supply line. The temperature probes were connected to a Cole Parmer Thermistor Thermometer. An ambient probe also was connected to the Thermistor Thermometer. The Arctic Sun was operated in Manual Mode with a water target of 4°C and 42°C. The ambient heat loss from the fluid delivery line was computed as:

$$ {\hbox{h}}_{\rm{l}} \times {\hbox{A}}_{\rm{l}} {\hbox{}}\;{\hbox{ = }}\;{\hbox{ 60 }} \times {\hbox{ }}\sigma {\hbox{ }}\times {\hbox{ Q }} \times {\hbox{ Cp }} \times {\hbox{(T}}_{\rm{r}} -{\hbox{ T}}_{\rm{s}} {\hbox{) /(T}}_{\rm{r}}-{\hbox{T}}_{\rm{a}} {\hbox{)}} $$

where h_l × A_l = heat transfer coefficient normalized for surface area (kcal/h(C) and T_a = ambient temperature (^oC). The value of the fluid delivery line heat loss was:

$$ {\hbox{h}}_{\rm{l}} \times {\hbox{A}}_{\rm{l}} {\hbox{ }}\;{\hbox{ = }}\;{\hbox{ 3}}{\hbox{.7 kcal/h}}^\circ {\hbox{C}} $$

The energy loss from the Energy Transfer Pads was determined on two volunteers that had a full set of five Energy Transfer Pads applied. Heat loss was measured using Concept Engineering Heat Flow Sensors and was recorded directly in kcal/m² mV. The subjects were laying supine in a hospital bed and were covered with a light blanket. Heat flux transducers were placed on pads covering the thighs, back, and abdomen. Measurements were area averaged and averaged for the two subjects to give a pad ambient heat transfer coefficient of:

$$ {\hbox{h}}_{\rm{p}} \times {\hbox{A}}_{\rm{p}} {\hbox{}}\;{\hbox{ = }}\;{\hbox{ q}}_{\rm{p}} {\hbox{/(T}}_{\rm{s}}-{\hbox{T}}_{\rm{a}} {\hbox{) = 6}}{\hbox{.1 kcal/h}}^\circ{\hbox{C}} $$

where q_p = pad heat loss (kcal/h).

The corrected heat transfer rate was computed as:

$$ {\hbox{q}}_{\rm{c}} \times {\hbox{q}}_{\rm{u}} {\hbox{ }}\;{\hbox{ = }}\;{\hbox{ (h}}_{\rm{l}} \times {\hbox{A}}_{\rm{l}} {\hbox{ + h}}_{\rm{p}} \times {\hbox{A}}_{\rm{p}} \times {\hbox{(T}}_{\rm{s}} -{\hbox{T}}_{\rm{a}} {\hbox{) }} $$

The total energy removed from a patient was computed as the integral of the corrected heat transfer rate over time. In equation form:

$$ {\hbox{Comulative Energy Transfer = }}\sum {\hbox{q}}_{{\rm{c,i}}} \times \Delta {\hbox{t}}_{\rm{i}} {\hbox{ }} $$

where q_c,i = corrected heat transfer rate (kcal/min) and Δt_i = time interval ( = 1 min).